early predictors of seizure outcome in …...early predictors of seizure outcome in newly diagnosed...
TRANSCRIPT
EARLY PREDICTORS OF SEIZURE OUTCOME IN NEWLY DIAGNOSED EPILEPSY
A Systematic Review of Prognosis Studies
Folarin Oluseye ABIMBOLA, MBChB
This thesis is presented for the degree of Master of Philosophy
School of Public Health, Sydney Medical School
Neurological and Mental Health Division The George Institute for Global Health
The University of Sydney
August 2010
1
TABLE OF CONTENTS DECLARATION…………………………………………………………………………………………………………………….….7 ACKNOWLEDGEMENT……………………………………………………………………………………………………….…..8 ABSTRACT……………………………………………………………………………………………………………………………...9
1 CHAPTER ONE: INTRODUCTION AND BACKGROUND ............................................................ 12
1.1 INTRODUCTION ................................................................................................................... 12
1.1.1 DEFINITION AND EPIDEMIOLOGY OF EPILEPSY ............................................................ 12
1.1.2 SEIZURES IN EPILEPSY ................................................................................................... 13
1.1.3 AETIOLOGY OF SEIZURES IN EPILEPSY .......................................................................... 16
1.1.4 CHALLENGES OF STUDYING THE PROGNOSIS OF EPILEPSY .......................................... 16
1.1.5 OUTCOMES IN EPILEPSY ............................................................................................... 18
1.1.6 ORGANISATION OF THE THESIS .................................................................................... 19
1.2 BACKGROUND TO PROGNOSIS STUDIES IN EPILEPSY ......................................................... 21
1.2.1 THE NATURAL HISTORY OF EPILEPSY ............................................................................ 21
1.2.2 WHEN TO COMMENCE OR DISCONTINUE MEDICATION ............................................. 25
1.2.3 WHEN TO CONSIDER OTHER INTERVENTIONS ............................................................. 27
1.2.4 THE SEIZURE OUTCOME OF EPILEPSY........................................................................... 33
1.3 SYSTEMATIC REVIEWS OF PROGNOSIS STUDIES ................................................................. 36
1.3.1 THE ROLE OF SYSTEMATIC REVIEWS ............................................................................ 36
1.3.2 HISTORY OF RESEARCH SYNTHESIS IN MEDICINE ......................................................... 37
1.3.3 CRITICISMS OF RESEARCH SYNTHESIS IN MEDICINE .................................................... 38
1.3.4 SYSTEMATIC REVIEW OF OBSERVATIONAL STUDIES .................................................... 38
1.3.5 SYSTEMATIC REVIEWS OF PROGNOSIS STUDIES IN EPILEPSY ...................................... 39
2 CHAPTER TWO: THE AIMS AND SIGNIFICANCE OF THIS THESIS ............................................ 42
3 CHAPTER THREE: METHODS .................................................................................................. 45
3.1 ELIGIBILITY CRITERIA............................................................................................................ 45
3.2 SEARCH STRATEGY ............................................................................................................... 46
3.2.1 MEDLINE ....................................................................................................................... 46
3.2.2 EMBASE ........................................................................................................................ 47
3.2.3 SCREENING THE RESULTS FOR ELIGIBLE PUBLICATIONS .............................................. 47
3.3 DATA EXTRACTION .............................................................................................................. 48
3.4 QUALITY APPRAISAL ............................................................................................................ 49
3.4.1 STUDY PARTICIPATION ................................................................................................. 49
3.4.2 STUDY ATTRITION ......................................................................................................... 50
3.4.3 PROGNOSTIC FACTOR MEASUREMENT ........................................................................ 52
2
3.4.4 OUTCOME MEASUREMENT .......................................................................................... 53
3.4.5 MULTIVARIATE REGRESSION ANALYSIS ........................................................................ 54
3.4.6 SUMMARY OF POTENTIAL FOR BIAS IN STUDIES ......................................................... 55
3.5 SYNTHESIS OF RESULTS ....................................................................................................... 55
3.6 EXTERNALLY VALIDATED PROGNOSTIC MODELS ................................................................ 57
3.6.1 CALIBRATION AND DISCRIMINATION ........................................................................... 57
3.6.2 INTERNAL VALIDATION ................................................................................................. 58
3.6.3 EXTERNAL VALIDATION ................................................................................................ 59
3.6.4 SENSIBILITY ................................................................................................................... 59
3.6.5 EFFECTS OF USE ON CLINICAL PRACTICE ...................................................................... 59
4 CHAPTER FOUR: RESULTS ...................................................................................................... 61
4.1 IDENTIFYING ELIGIBLE PUBLICATIONS ................................................................................. 61
4.2 DESCRIPTIVE CHARACTERISTICS OF ELIGIBLE PUBLICATIONS ............................................. 63
4.3 REPORTING CHARACTERISTICS OF ELIGIBLE PUBLICATIONS ............................................... 68
4.4 QUALITY APPRAISAL OF ELIGIBLE PUBLICATIONS ............................................................... 70
4.4.1 STUDY PARTICIPATION ................................................................................................. 70
4.4.2 STUDY ATTRITION ......................................................................................................... 70
4.4.3 PROGNOSTIC FACTORS ................................................................................................. 71
4.4.4 OUTCOME MEASUREMENT .......................................................................................... 71
4.4.5 MULTIVARIATE ANALYSIS ............................................................................................. 71
4.5 STUDY CATEGORIES BY OUTCOME MEASURE ..................................................................... 73
4.6 REVIEW OF STUDIES BY OUTCOME CATEGORY ................................................................... 76
5 CHAPTER FIVE: DISCUSSSION .............................................................................................. 104
5.1 THE LITERATURE SEARCH .................................................................................................. 104
5.2 QUALITY AND REPORTING CHARACTERISTICS .................................................................. 105
5.3 THE CLASSIFICATION OF STUDIES ...................................................................................... 106
5.4 EARLY/IMMEDIATE REMISSION ......................................................................................... 107
5.5 REMISSION OFF ANTIEPILEPTIC MEDICATION ................................................................... 109
5.6 REMISSION ON OR OFF MEDICATION ............................................................................... 112
5.7 INTRACTABILITY ................................................................................................................. 117
5.8 REMISSION AFTER RELAPSE ............................................................................................... 119
5.9 IMPLICATIONS FOR CLINICAL PRACTICE ............................................................................ 119
5.10 DIRECTIONS FOR FUTURE RESEARCH .............................................................................. 120
5.11 STRENGTHS AND WEAKNESSES OF THE REVIEW ............................................................ 122
6 CHAPTER SIX: CONCLUSION ................................................................................................ 125
3
7 APPENDICES ......................................................................................................................... 129
7.1 Appendix I: PROTOCOL ...................................................................................................... 129
7.2 Appendix II: LIST OF CITATIONS ......................................................................................... 136
7.3 Appendix III: DATA EXTRACTION AND QUALITY APPRAISAL FORM .................................. 141
7.4 Appendix IV: ITEMS FOR THE ASSESSSMENT OF BIAS IN PROGNOSIS STUDIES ................ 149
7.5 Appendix V: ITEMS TO BE CONSIDERED IN REPORTING OBSERVATIONAL STUDIES ......... 150
8 REFERENCES ........................................................................................................................ 152
4
LIST OF TABLES Table 1 | Eligible Publications by Year, Journal and Country ........................................................ 64 Table 2 | Descriptive Characteristics of Cohorts, Publications and Studies (European Cohorts) . 66 Table 3 | Descriptive Characteristics of Cohorts, Publications and Studies (Other Cohorts) ....... 67 Table 4 | Reporting characteristics of publications with multivariate analysis ............................ 69 Table 5 | Details of Quality Appraisal of Publications ................................................................... 72 Table 6 | Studies predicting immediate remission with information on potential for bias ......... 78 Table 7 | Consistent predictors and non-predictors of immediate remission .............................. 79 Table 8 | Studies predicting remission off AED (Nova Scotia and DSEC Cohort) .......................... 81 Table 9 | Studies predicting remission off AED (Montreal and Turku Cohort) ............................. 82 Table 10 | Consistent predictors and non-predictors of remission off AED ................................. 83 Table 11 | Externally validated models predicting remission off AED .......................................... 85 Table 12 | Studies predicting remission on or off AED (the DSEC Cohort) ................................... 88 Table 13 | Studies predicting remission on or off AED (Others 1) ................................................ 89 Table 14 | Studies predicting remission on or off AED (Others 2) ................................................ 90 Table 15 | Consistent predictors and non-predictors of remission on or off AED (1) .................. 91 Table 16 | Consistent predictors and non-predictors of remission on or off AED (2) .................. 92 Table 17 | Externally validated models predicting remission on or off antiepileptic drugs ......... 94 Table 18 | Studies predicting intractability (full cohort) ............................................................... 96 Table 19 | Studies predicting intractability (nested case control) ................................................ 97 Table 20 | Studies predicting intractability (nested case control by Ko et al) .............................. 98 Table 21 | Consistent predictors and non-predictors of intractability ......................................... 99 Table 22 | Studies predicting no remission after relapse ........................................................... 102 Table 23 | Predictors and non-predictors of no remission after relapse .................................... 102 Table 24 | Consistent early predictors of seizure outcome in newly diagnosed epilepsy .......... 126
5
LIST OF FIGURES Figure 1 | Pathophysiology of Focal Seizures ................................................................................ 13 Figure 2 | Pathophysiology of Primary Generalised Seizures ....................................................... 14 Figure 3 | Pathophysiology of Secondary Generalisation of Seizures........................................... 14 Figure 4 | Natural History of Newly Diagnosed Epilepsy .............................................................. 33 Figure 5 | Natural History of Newly Diagnosed Epilepsy (with additional granularity) ................ 34 Figure 6 | Flow diagram of the systematic review ........................................................................ 62 Figure 7 | Type I studies (Predicting Immediate Remission) ......................................................... 73 Figure 8 | Type II studies (Predicting Remission off AED) ............................................................. 74 Figure 9 | Type III studies (Predicting Remission on or off AED) .................................................. 74 Figure 10 | Type IV studies (Predicting Intractability including the entire cohort)....................... 75 Figure 11 | Type IV studies (Predicting Intractability in nested case control studies) .................. 75 Figure 12 | Type V studies (Predicting no remission after relapse) .............................................. 76
6
LIST OF BOXES Box 1 | Relative risk (RR) of achieving early remission in patients with two or more seizures before intake ................................................................................................................................. 77 Box 2 | Relative risk (RR) of achieving early remission in patients with remote symptomatic aetiology ........................................................................................................................................ 77 Box 3 | Relative risk (RR) of remission off AED in children having more than 1 seizure 6 to 12 months on AED .............................................................................................................................. 80 Box 4 | Relative risk (RR) of remission off AED in children with intellectual disability (models with only onset variables) ..................................................................................................................... 80 Box 5 | Relative risk (RR) of remission off AED in children with intellectual disability (models combining onset and 1 year variables).......................................................................................... 83 Box 6 | Risk of achieving remission in patients with mixed seizure types at onset ...................... 86 Box 7 | Relative risk (RR) of achieving remission in patients who have remote symptomatic aetiology ........................................................................................................................................ 87 Box 8 | Odds ratio (OR) of achieving remission in patients with intellectual disability ................ 87 Box 9 | Relative risk (RR) of achieving remission in patients as a function of Log N of Seizures before index .................................................................................................................................. 87 Box 10 | Relative risk of achieving remission in patients as a function of Log N of Seizures from index to 6 months .......................................................................................................................... 93 Box 11 | Relative risk (RR) of having intractable seizures for each increasing year of age at onset of seizures ...................................................................................................................................... 95 Box 12 | Relative risk (RR) of having intractable seizures with onset of seizures in infancy (age <1 year) ............................................................................................................................................. 100 Box 13 | Relative risk (RR) of having intractable seizures in patients with remote symptomatic aetiology ...................................................................................................................................... 100 Box 14 | Relative risk (RR) of having intractable seizures in patients with idiopathic aetiology 100 Box 15 | Odds ratio (OR) of having intractable seizures in patients with intellectual disability . 100 Box 16 | Proportion of patients in immediate remission at 2 years by the proportion of patients on AED ......................................................................................................................................... 107 Box 17 | Proportion of patients in remission off AED arranged according to potential for bias 110 Box 18 |Remission as seizure-free period immediately before last assessment i.e. Terminal Remission (TR) ............................................................................................................................. 113 Box 19 | Remission (R) as seizure-free period during follow up ................................................. 114 Box 20 | Influence of definition and study design/setting on the proportion of patients with intractable seizures ..................................................................................................................... 118
7
DECLARATION I hereby declare that the work described herein is original and my own, except where due
acknowledgement has been made to the work of others. This thesis was completed while
enrolled in the School of Public Health at the University of Sydney, under the superv ision of
Dr Alexandra Martiniuk and Professor Craig Anderson based at the George Institute for
Global Health in Sydney who provided invaluable advice and many suggestions for the
improvement of the study and this document. I certify that this thesis has not already been
submitted, wholly or in part, for a higher degree to any other university or institution. All
the help received in preparing this thesis, all material from published sources, and all
material copied from the work of others have been properly acknowledged and referenced.
_______________________________
Folarin Oluseye (Seye) ABIMBOLA
4 August 2010
8
ACKNOWLEGEMENT
I wish I could say like Mark Twain that if only I had more time, my thesis would have been much
shorter than this; it is not in the nature of theses to be decidedly brief. I therefore express deep
gratitude to my supervisors who took the pain and patience to read through and annotate with
comments, chicken scratches and smiley faces the overly chatty and journalistic ramblings that
made up the early drafts of the thesis.
In particular, thanks to Maree Hackett for providing, early on before the study had a life of its
own, papers, two of which were hers, that allowed me to think clearly about what I wanted the
thesis to achieve and how I might set out to report the systematic review; to Nick Glozier for
helping to plug several holes in my thinking; to Alexandra Martiniuk for her trust and confidence
which allowed me to think freely without crippling restrictions and overwhelming self-doubt;
and finally to Craig Anderson for several, mostly unspoken acts of kindness and encouragement.
I especially acknowledge the support of Alexandra Martiniuk and Craig Anderson who ensured
my move to the George Institute for Global Health and the University of Sydney, and the aid of
the Rotary Foundation Ambassadorial Scholarship from Rotary International for which I am
eternally grateful. My life in Sydney has been enriched by the generous friendship of the family
of Jeremy Wright and Marilyn Smith, including their daughter Sophie Wright and especially their
son, Chris Wright. Theirs has been a home away from home. I am most grateful for their love.
Thanks to Kathleen Bongiovanni who volunteered to join in the week-long 9-5 screening of the
EndNote library. Thanks to the people who ensured that my lesser brain received nourishment
with poetry, pep talk and the sheer pleasure of spontaneous company: Ulli and Georgina Beier,
Pip Crooks, Kristine Maddock, Lara Laloko, and Temi Giwa. Thanks to those who were obliging
and enthusiastic sounding boards when I was starved for talk: Chioma Onukwue (who also
proof-read the entire thesis), and Andre Pascal Kengne. Thanks to my friends whose faith in me
is a great spur: Tunde Ademola, Remi Oyedeji, Tope Ojo, Wumi Adesida, Femi Owagbemi,
Olumide Ilesanmi; and thanks especially to Oine Omakwu for love, and “the peace that passes
all understanding.”
Thanks to my wonderful siblings Olumide, Opeyemi and Oluwatoyin who graciously pretended
to overlook my career choice to be an academic, and to earn hardly anything for a while. Thanks
to my father, who although may not understand what I am doing with my life, has carefully
remained quiet about it. Finally, thanks to my very dear mother who passed away when this
thesis was just about to be formed: a child’s passion became her own passion and she gave
willingly, spontaneously and with unflagging commitment. I dedicate the work to her memory
and to my infant niece, Zoe Yetunde, born only 24 days after my mother passed away, and who
would have been her much anticipated first grandchild.
9
ABSTRACT
The physical and psychosocial consequences of seizures in epilepsy are more prevalent in people
with uncontrolled seizures, and are lessened with seizure control. Early management strategies
and patient counselling may be informed by the predictors of control in newly diagnosed
patients. The main objective of this systematic review was to identify variables that consistently
and independently predict seizure outcome in people with newly diagnosed epilepsy.
English-language publications identified from electronic databases (MEDLINE and EMBASE) and
reference lists of included studies, published until March 2010 were reviewed. The publications
were of nested case control and cohort studies of unselected population of at least 100 people
with epilepsy with multivariate analysis of the effect on seizure outcome of predictor variables
collected within the first year of diagnosis in patients followed up for at least 1 year.
The quality of included studies was appraised for their likelihood for bias in five areas: study
participation, study attrition, prognostic factor measurement, outcome measurement and
statistical analysis. Data from each study were independently extracted three times; at each
succeeding stage, the extracted data were compared with the previous extraction and where
there were discrepancies, clarification made by consulting the publication. Consistent predictors
were identified in more than 1 study from different cohorts.
There were 52 studies from a total of 33 publications. Five studies predicted immediate
remission of seizures; 9 studies predicted remission off antiepileptic medication; 20 studies
predicted remission on or off antiepileptic medication; 12 studies predicted intractability to
antiepileptic medication and 1 study developed a model to predict not achieving remission after
1 or more relapses following an initial period of remission. 5 studies externally validated models
predicting seizure outcome.
Two factors reduce the chance of achieving immediate remission: More than 1 seizure before
recruitment [RR 0.63 (95%CI 0.36-1.11)] and remote symptomatic aetiology [childhood-onset
epilepsy: RR 0.59 (95%CI 0.41-0.86), adult-onset epilepsy: RR 0.44 (95% CI 0.26-0.77)].
Having more than 1 seizure in the period between 6 and 12 months on medication [RR 0.24
(95%CI 0.10-0.60)] and intellectual disability [RR 0.77 (95%CI 0.61-0.94)] reduce the chance of
achieving remission off medication in childhood-onset epilepsy. None of the studies predicting
remission off medication were of adult-onset epilepsy.
Five factors reduce the chance of achieving remission on or off medication are: More than 1
seizure before diagnosis [childhood-onset epilepsy: RR 0.66 (95%CI 0.46-0.95), adult-onset-
epilepsy: 0.81 (95%CI 0.66-0.99)] for the natural logarithm of every additional seizure; seizures
in the first 6 months after the index seizure [childhood-onset epilepsy: RR 0.50 (95%CI 0.35-
10
0.72), adult-onset-epilepsy: RR 0.59 (95%CI 0.50-0.70) for the natural logarithm of every
additional seizure; mixed seizure types at onset [RR 0.70 (95%CI 0.37-1.04), adult-onset epilepsy:
OR 0.23 (95% CI 0.10-0.48)]; intellectual disability [childhood-onset epilepsy: OR 0.40 (95%CI
0.18-0.90), adult-onset epilepsy: OR 0.10 (95%CI 0.05-0.25)]; and remote symptomatic aetiology
[childhood-onset epilepsy: RR 0.63 (95% CI 0.47-0.84), adult-onset epilepsy: RR 0.44 (95% CI not
reported)].
Onset of seizures in infancy [RR 5.48 (95%CI 2.10-9.64)], intellectual disability [OR 18.2 (95%CI
5.2-63.6)], and remote symptomatic aetiology [RR 5.48 (95%CI 2.10-9.64)] increase the risk of
medical intractability, while idiopathic aetiology [RR 0.20 (95%CI 0.0-0.80)] reduces the risk of
intractability in childhood-onset epilepsy. None of the studies predicting medical intractability
were of adult-onset epilepsy.
The externally validated models have little predictive gain over information provided by simple
prevalence rates of seizure outcome, and the models predict wrongly in about 1 out of 3
children in the development and external validation cohorts. None of the models developed in
adult-onset epilepsy were externally validated.
The study suggests that onset of seizures in infancy, number of seizures (before diagnosis, in the
first 6 months after diagnosis, and between 6 and 12 months on medication), intellectual
disability and the aetiology of seizures are the important predictors of seizure outcome in newly
diagnosed epilepsy. The study demonstrates the feasibility of systematic review with thorough
quality appraisal as a means of identifying the consistent predictors of an outcome in
exploratory prognostic factor studies. The review also shows the need for further studies of the
prognosis of adult-onset epilepsy.
11
12
1 CHAPTER ONE: INTRODUCTION AND BACKGROUND
This chapter is in 3 parts. The first part introduces and provides an overview of the main thrust
of the thesis. The second part presents a general background to prognosis studies in epilepsy
and describes the benefits of early determination of the prognosis of epilepsy in newly
diagnosed patients. The second part also presents a framework for classifying the different
potential prognostic categories of a newly diagnosed patient with epilepsy. The third part is an
overview of the role of systematic review in medicine with particular application to prognosis
studies in epilepsy.
1.1 INTRODUCTION
1.1.1 DEFINITION AND EPIDEMIOLOGY OF EPILEPSY Epilepsy is a chronic neurological disorder characterised by a recurrent tendency to have
spontaneous, intermittent, abnormal electrical activity in a part of the brain, which manifest as
seizures, and diagnosed as the result of a patient having a second unprovoked seizure, with at
least 24 hours between the first and second seizure. This definition regards an episode of status
epilepticus (a seizure lasting more than 30 minutes or repeated seizures without intervening
periods of regained function or consciousness) as a single seizure.(1)
However, the definition of epilepsy as a tendency to have recurrent seizures excludes seizures
that are provoked (i.e. not spontaneous, therefore “acute symptomatic”) by an obvious and
immediate preceding cause e.g. an acute systemic or metabolic imbalance, drugs or toxins, or a
recent cerebral damage from stroke, trauma or infection. Seizures occurring in children between
6 months and 6 years only within the context of a febrile illness without the evidence of
intracranial aetiology (febrile seizures) are also excluded, as are seizures occurring only within
the neonatal period.(1, 2)
The World Health Organisation (WHO) estimates the point prevalence of active epilepsy (i.e.
people with continuing seizures or the need for treatment) as generally 4 to 10 per 1,000
people, and in developing countries from 6 to 10 per 1,000. It is also estimated that at least 50
million people in the world have epilepsy as 43.7 million people were reported to have epilepsy
from 108 countries covering 85.4% of the world in a WHO survey.(3, 4) The mean number of
people with epilepsy per 1000 population is 8.93. This varies from 7.99 in high-income countries
to 9.50 in low-income countries.(3) However, the incidence of epilepsy in developing countries is
about twice that in developed countries, and the WHO estimates that about 80% of the world’s
epilepsy patients are in developing countries.(3)
13
In developed countries, new-onset seizure occurs in approximately 80 people per 100,000 in a
year. The estimates of annual incidence of epilepsy in the general population range from 30 to
57 per 100,000.(5) However, the incidence rate of epilepsy has a bimodal pattern in relation to
age: it is high in the paediatric population (about half of all epilepsy cases are diagnosed in
childhood or adolescence), decreasing through adulthood until approximately age 60, when the
incidence again begins to increase.(6, 7) With the onset of epilepsy, a chronic disorder, being
common in childhood and adolescence, and due to the burden of experiencing unpredictable
paroxysmal seizure events and its psychosocial implications, epilepsy contributes about 1% of
the global burden of disease.(4)
1.1.2 SEIZURES IN EPILEPSY A working knowledge of basic clinical epilepsy is assumed in this thesis. However, the following
account of the clinical and pathophysiological foundations of seizures in epilepsy is based on a
2009 monograph on clinical epilepsy by Shorvon,(8) updated with the use of terminologies for
seizure classification from the 2010 report of the ILAE (International League Against Epilepsy)
Commission on Classification and Terminology.(9)
The term “seizure” refers to the transient clinical manifestation of an episodic, abnormal,
excessive, hypersynchronous discharge of a population of epileptic cortical neurons. For a
particular patient, the seizure tends to be stereotyped, although it may take many forms (mixed
seizure types) in others. The form seizures in epilepsy take could be conceptualised according to
how it is experienced by the patient (as a motor, somatosensory, autonomic or psychic
manifestation with or without accompanying impaired level of consciousness) or more
conveniently, on physiological grounds, as either focal or generalised, in relation to how seizure
activity originates within the brain.
The terms “focal” and “generalised” have been used to express a dichotomous classification for
seizures and epilepsy, although they do not represent a clear dichotomy in the pathophysiology
of seizures.(9) In focal seizures, the paroxysmal neuronal activity giving rise to the seizure is
limited to a focal area in one half of the cerebrum, with consciousness preserved, impaired or
completely lost. (Figure 1)
Figure 1 | Pathophysiology of Focal Seizures
[Image from Dekker 2002(10)]
14
For generalised seizures, the electrophysiological abnormality involves large areas of the 2
cerebral hemispheres simultaneously and synchronously. These generalised seizures are always
accompanied by at least impaired level of consciousness. (Figure 2)
Figure 2 | Pathophysiology of Primary Generalised Seizures
[Image from Dekker 2002(10)]
The third category that blurs the distinction between focal and generalised seizures is the
secondarily generalised seizures in which the initially focal neuronal discharge spreads from 1
hemisphere to the other to become generalised. (Figure 3)
Figure 3 | Pathophysiology of Secondary Generalisation of Seizures
[Image from Dekker 2002(10)]
FOCAL SEIZURES Focal seizures that remain localised manifest symptoms depending on the area of the cerebral
cortex affected. In about 15% of patients with epilepsy the only seizure type they experience are
those that start focally and remain focal with consciousness preserved, impaired or completely
lost.
Focal seizures with motor manifestations occur when seizure focus is in the precentral gyrus
(motor cortex) affecting the contralateral face, arm, trunk or leg characterised by rhythmical
jerking or sustained spasm. The focal seizures with somatosensory or special sensory (simple
hallucinations) manifestations arise in the sensory cortex, and may cause unpleasant tingling or
15
electric sensations in the contralateral face and/or limb. Focal seizures with sensory
manifestations may originate from the visual (occipital) cortex with hallucination such as balls of
light or patterns of colour, or from the temporal lobe, where hallucination may involve faces or
scenes. Focal seizures with autonomic manifestations include changes in skin colour, blood
pressure, heart rate, pupil size and piloerection.
Focal seizures with psychic manifestations occur especially when the focus is in the temporal
lobe, with 6 categories of manifestation: 1.) Dysphasia when cortical speech areas of the frontal
and temporoparietal lobes are affected; 2.) Dysmnesia, in which memory problems such as déjà
vu, jamais vu, flashbacks or panoramic experiences; 3.) Cognitive, where dreamy states and
sensations of unreality or depersonalisation is part of the seizure; 4.) Affective, in which fear
occurs commonly, but depression, anger and irritability may also be a feature of the seizure; 5.)
Illusions, which may be of size, weight, shape, distance or sound which occurs usually with a
temporal or parieto-occipital focus; and 6.) Hallucinations, which may be visual, auditory,
gustatory or olfactory usually due to temporal or parieto-occipital focus.
GENERALISED SEIZURES Generalised seizures may be primary or secondarily generalised. In about 25% of patients with
epilepsy, only primary generalised seizure occurs. There are 6 categories of primary generalised
seizure, all accompanied with impaired consciousness from the onset of seizure activity as
seizure originates extensively within cerebral cortex and subcortical structures.
However, 20% of all patients with epilepsy have primarily generalised seizures with only tonic
clonic manifestation where the patient suddenly becomes rigid (tonus), then unconscious, falls if
standing, and respiration is arrested with possible central cyanosis. The rigidity then becomes
periodically relaxed (clonus). The patient subsequently enters a flaccid state of deep coma.
When consciousness is regained, there is first a phase of confusion and disorientation following
which the patient regains full memory, feeling terrible, with headache, and sleepiness. There is
possibility of loss of continence occurring as well as tongue biting during the seizure.
The other types of primary generalised seizures make up the remaining 5% of patients with
epilepsy having only primary generalised seizures: 1.) Absence Seizures occur typically as an
abrupt loss of consciousness, but with preservation of muscle tone, and in its atypical form, loss
of consciousness is not complete, with varying degrees of loss of muscle tone. 2.) Myoclonic
Seizures present as brief contraction of a muscle or muscle group or several muscle groups,
which may be single or repetitive and with varying severity from a twitch to severe jerking. 3.)
Clonic Seizures are often asymmetric with irregular clonic jerking, while 4.) Tonic Seizures,
present as tonic muscle contraction without a clonic (jerking) phase and 5.) Atonic Seizures, in its
most severe form, manifests as a sudden complete loss of postural tone and the patient drops
or, it may be more restricted, resulting in loss of tone in specific groups of muscle.
16
These seizures types are primary generalised. However, they may also be secondary generalised,
when the seizure starts with an aura (a partial seizure). These secondary generalised seizures
occur exclusively in about 60% of patients with epilepsy.
1.1.3 AETIOLOGY OF SEIZURES IN EPILEPSY The concept for assessing, categorising and reporting on the underlying cause of epilepsy, was
described in the 1989 ILAE Commission on Classification and Terminology report as being
idiopathic, symptomatic, or cryptogenic.(11) In idiopathic aetiology, “there is no underlying
cause other than a possible hereditary predisposition… and a presumed genetic aetiology.’’
Symptomatic aetiology is “considered the consequence of a known or suspected disorder of the
central nervous system (CNS). The term cryptogenic “refers to a disorder whose cause is hidden
or occult… presumed to be symptomatic, but the aetiology is not known.”(11)
Many human beings and other animals will have seizures in abnormal metabolic circumstances
such as hyponatraemia, hypocalcaemia, hypoxia, hypo or hyperglycaemia.(5) However, the
tendency to have recurrent seizures is not common to all and there are aetiological factors
responsible, even though these factors cannot be identified in about 30% of patients
(cryptogenic). The cause is genetic in about another 30% (idiopathic) and symptomatic (or
remote symptomatic, to distinguish from acute symptomatic seizures that are excluded from
the definition of epilepsy) in about 40% due to cerebrovascular disease (10-20%), malformations
including hippocampal sclerosis, other neurological disorders and neurodegenerative disease
(10-20%) intracranial tumour (5-10%), brain trauma (5%), cerebral infection (5%), or metabolic
disorder and toxins including alcohol (5%).
The more recent 2010 report of the 2005-2009 ILAE Commission on Classification and
Terminology (9) recommends more clearly defined and less confusing and conflating terms and
concepts in place of the traditional terminologies. The authors of the report propose that
“Genetic” replaces Idiopathic, “Structural/Metabolic” replaces symptomatic, and “Unknown
Cause” replaces cryptogenic. The only major difference in the proposed use and definition of
these terms relates specifically to advances in the use of genetics to better classify the aetiology
of epilepsy. This thesis keeps to terminologies in current use.
1.1.4 CHALLENGES OF STUDYING THE PROGNOSIS OF EPILEPSY
DIAGNOSIS AND CASE ASCERTAINMENT Epilepsy syndromes (epileptic seizures characterised by a consistent cluster of patient
characteristics, with signs and symptoms occurring together) are organised with a complex
relationship to seizure pathophysiology and aetiology.(8) The ILAE classification of epilepsy
syndromes follows the pattern of aetiological classification as syndromes of epilepsy arising
from focal seizures are grouped under the headings of idiopathic, symptomatic and cryptogenic
17
aetiology as are those arising from generalised seizures and those that are not classified either
as focal or generalised. The fourth category under which epilepsy syndromes are grouped is that
of “situation-related seizures” but without aetiological subdivisions. Exploration of these
syndromes is however beyond the scope of this thesis.
Epilepsy, like many clinical constructs is not a singular disease entity, but a manifestation of
underlying disease processes. The attempt to bring the different entities with a common
tendency to have repeated seizures, albeit with a myriad of possible seizure types and aetiology
and syndromes beneath one umbrella is therefore fraught with the contentious issue of
classification and terminology.(12-17) This has led to repeated attempts by the International
League Against Epilepsy to standardise definitions, terminologies and classification to provide a
reproducible conceptual framework for organising and differentiating epileptic seizures and
types of epilepsy and the standard indices for epidemiological measures of prevalence and
incidence.(1, 2, 9, 11, 18)
Epilepsy is essentially a clinical diagnosis, based on eyewitness account or video recording of the
seizure, as clinical examination and investigations may be normal between seizures. Patients
may also not be aware of the nature of the seizure, and seizures may go unnoticed. These lead
to seizures going unreported. There may also be denial of occurrence of seizures in order to
avoid the stigma attached to a diagnosis of epilepsy. It is also possible that patients with
infrequent seizures or mild epilepsy may not seek medical attention.(19, 20)
These factors result in difficulty with adequate case identification and case ascertainment, in
determining the specific seizure type, the aetiology, and therefore the specific syndromic
diagnosis of each patient’s epilepsy; this extends also to the impact on epidemiological studies
on populations of patients with epilepsy.(1, 2, 14, 20) However, when such syndromic diagnosis
is made, the small number of patients within each group precludes much statistical analysis
based on each of these groups as several epilepsy syndromes are rare.(21) Hence, much of the
epidemiological studies in epilepsy do not classify patients beyond the major seizure type and
aetiological classifications.
STUDY DESIGN The retrospective study design as well as hospital based studies tend to lead to bias in selecting
for patients with more severe disease manifestation and course.(20, 22) Therefore prospective
case ascertainment and follow up is preferred as it reduces the risk for selection bias and also
allows for optimal assessment of predictors and outcome.(23)However, the difficulty and cost of
ensuring prospective case ascertainment in a population based study makes the study of
epidemiology and especially prognosis of epilepsy more difficult and tasking in terms of cost and
logistics. Ideally, to determine the prognosis of epilepsy in a group of patients, they have to be
assembled at a comparable stage in their disease progression. This is at the point where they
18
could be described as newly diagnosed, and better still, for each patient, at the exact point of
having a second seizure, which diagnoses epilepsy.(23, 24)
MULTIVARIATE MODELS To identify independent predictors of the outcome of newly diagnosed epilepsy, by avoiding the
effect of confounders, it is imperative that multivariate analysis is conducted within a population
of people with epilepsy.(23) However, the prognosis of epilepsy is dynamic:(25) some patients
have a remitting course, some have a remitting-relapsing course and others a worsening course.
Therefore, a multivariate analysis conducted on such a dynamic population of patients is as
described by Iezzoni(26) only “a snapshot in place and time, not fundamental truth.”
This thesis investigates the early predictors of the prognosis of epilepsy in newly diagnosed
patients. It is a systematic review of studies that have used multivariate regression models to
identify independent predictors of outcome in people with newly diagnosed epilepsy. The
patients within these studies may be at different stages of the course of their epilepsy. Thus, the
result of a multivariate analysis that identifies independent predictors may only be a conceptual
representation, and not necessarily a true reflection of the prognosis of epilepsy.
For a model to be useful in another population, it has to be confirmed as generalisable in an
external validation study of patients within another cohort. However, this external validation
only answers the Iezzoni’s question of place.(26, 27) The question of time in the course of the
epilepsy for individual patients within a particular cohort studied can only be satisfied if all the
patients in the cohort were recruited and followed at the same point in the course of the
epilepsy.(27, 28) Although difficult to achieve, the point of a patient’s second seizure may mark
a gold standard for recruitment and commencement of follow up in prospective studies of the
prognosis of epilepsy.(24)
1.1.5 OUTCOMES IN EPILEPSY The focus of this thesis is restricted to the prognosis of epilepsy as it relates specifically to the
outcome of seizure, in terms of the continued occurrence or otherwise of seizure events
following the diagnosis of epilepsy and does not include circumstances or events that may result
from seizures: physical effects like fracture, burns and drowning from falls or loss of
consciousness, psychosocial outcome such as cognitive deficit, psychopathology (depression and
anxiety), dysfunctional relationships (overprotection and overdependence in relationships),
restriction in education, employment and leisure activity, and problems relating to low
confidence and self-esteem leading to underachievement and social withdrawal.(8, 29-31)
These outcomes may be a result of the unpredictability of seizures or as a result of perceived or
real stigma. They are however by no means less important. Indeed, these secondary
psychological and social limitations that epilepsy imposes are more an issue for some patients
than the occurrence of seizures.(31) The fear of death and death itself is another consequence
19
of epilepsy. The annual mortality in people with epilepsy is slightly higher than in the general
population and there is also relatively lower life expectancy. Death may result from physical
accidents resulting from seizures, status epilepticus, suicide,(32-34) or sudden unexpected
death in epilepsy (SUDEP) which occurs following generalised tonic clonic seizures, commonly
during sleep and possibly as a result of respiratory failure or cardiac arrhythmias.(35-37)
However, these potential consequences of seizures are reduced with control of seizures.(30, 31,
35-37) For example, the overall risk of SUDEP of 1 per 1000 per year, falls to zero with seizure
remission and rises to 1 per 100 per year in patients with frequent and severe tonic clonic
seizures.(8) This fact underscores the importance of this thesis in addressing the question of
identifying the early predictors of seizure outcome in patients with newly diagnosed epilepsy.
The results will inform management decisions that not only would help reduce the burden of
seizures but may also reduce the incidence of the consequences of continued seizures.
1.1.6 ORGANISATION OF THE THESIS The thesis is organised into 6 chapters and an appendix thus:
CHAPTER ONE Chapter 1 introduces and provides an overview of the main thrust of the thesis. The chapter also
presents a general background to prognosis studies in epilepsy and describes the benefits of
early determination of the prognosis of epilepsy in newly diagnosed patients. There is also a
discussion of the framework for classifying the different potential prognostic categories of a
newly diagnosed patient with epilepsy. The chapter ends with an overview of the role of
systematic review in medicine with particular application to prognosis studies in epilepsy.
CHAPTER TWO Chapter 2 presents the research questions that the thesis aims to answer. The chapter also
discusses the central contribution to the thesis to the systematic review and meta-analysis of
prognosis studies in general and particularly to studies of prognosis in epilepsy.
CHAPTER THREE Chapter 3 is a description of the systematic review process: eligibility criteria, the search
strategy, how each category of papers was excluded, and how studies from the same cohort
were handled. It also includes a discussion of the iterative process of developing the quality
appraisal and data extraction forms and how the results of each study was assessed for quality
i.e. tendency for bias. It also discusses the criteria for assessing the quality of externally
validated predictive models. The chapter also discusses how consistent predictors of outcome
were identified and the risk estimates compared across studies.
20
CHAPTER FOUR Chapter 4 presents the result of the systematic review. The chapter starts by reporting the
results of the process of identifying eligible publications, followed by a description of included
studies, how they relate to the cohorts from which they are derived and the publications that
report them, and an account of the reporting characteristics of the eligible publications. There is
a detailed report of the quality appraisal of studies included from each publication. The studies
were subsequently disaggregated according to the seizure outcome predicted, and each
category of study is further evaluated and appraised with a view to: 1) Explaining the similarities
and differences in results in relation to study characteristics and potential for bias, 2) Identifying
consistent predictors and non-predictors within each outcome category, and 3) Assessing the
quality and performance of externally validated models within each outcome category.
CHAPTER FIVE Chapter 5 is a general discussion of the results of the review. The chapter is presented according
to recommendations that advocate structured discussion of the results of scientific
research.(38) The principal findings of 1) the literature search; 2) the quality appraisal of studies
and reporting characteristics of publications; 3) the study classification; and 4) the studies within
each category, were stated and interpreted. Thereafter, possible explanations for the results are
presented and their implications for clinical practice are stated where applicable. There are also
suggestions for future studies and the direction of future research for each set of results. The
strengths of the study are highlighted, and the weaknesses are also discussed with a view to
making recommendations for future studies.
CHAPTER SIX Chapter 6 provides a summary of the key findings of the review and considers the results in the
context a previous review of predictors of outcome in patients with 1 seizure.
APPENDICES Five appendices were attached to this thesis. Appendix I is the protocol of the systematic
review; Appendix II is the complete list of 154 unique citations (from MEDLINE, EMBASE and
screening the references of eligible publications) with their origin indicated alongside the
corresponding reasons for inclusion and exclusion; Appendix III contains three versions of the
data extraction and quality appraisal form, from the first two pilot versions to the final version;
Appendix IV presents the items to be considered for assessment of bias in prognosis studies as
identified by Hayden et al(39); Appendix V presents the items to be considered in reporting
observational studies according to the STROBE (Strengthening the reporting of Observational
Studies) checklist.(40)
21
1.2 BACKGROUND TO PROGNOSIS STUDIES IN EPILEPSY This second part of the chapter discusses the prognosis of epilepsy and the importance of the
study of the prognosis in epilepsy.
There are different kinds of information that may arise from the study of prognosis. Fletcher et
al(41) identified 4 characteristics of disorders: response (evidence of improvement), remission
(disorder becomes undetectable), recurrence (return of the disorder after remission) and
duration (how long the disorder lasts) all of which are amenable to prognosis studies. These
studies also investigate the relationship between occurrences of outcomes and predictors in
defined populations of people with disease.(23, 42)
In addition, Windeler(43)made the observation of similarity between diagnosis and prognosis,
pointing out that the same factors may be responsible for diagnosis and prognosis. This
particularly applies to seizure outcome in epilepsy as diagnosis of epilepsy is confirmed on the
second unprovoked seizure. Therefore, while the study of seizure recurrence after the first
unprovoked seizure(44) is in effect a study of diagnosis, a prognosis study considers recurrence
or otherwise after a second unprovoked seizure.
The importance of prognostic research in epilepsy as in other chronic diseases is overwhelming.
Altman and Lyman (45) identified the uses of prognosis studies among which are to: “1.)
improve understanding of the disease process; 2.) define risk groups based on prognosis; 3.)
predict disease outcome more accurately or parsimoniously; and 4.) guide clinical decision
making, including treatment selection and patient counselling.”
1.2.1 THE NATURAL HISTORY OF EPILEPSY The natural history of epilepsy has important relevance for the better understanding of
underlying neurobiology of the disorders that result in epilepsy, in planning and evaluating
treatment strategies and health resource allocation. This segment discusses the natural history
of epilepsy first with historical considerations of knowledge and attitude towards the prognosis
of epilepsy, then the prognosis of epilepsy in patients who are not treated with anti-epileptic
medications and evidence from patient populations who are predominantly treated with
antiepileptic drugs with an emphasis on the dynamic nature of the course of epilepsy.
HISTORICAL CONSIDERATIONS There was an early documented hint at the natural history and prognosis of epilepsy by Plato
(c.429–347 BC) who wrote in Laws that whoever procures a slave with epilepsy had 1 year to
return the slave. For him, the prognosis of epilepsy might be ascertained within 1 year, as
against tuberculosis, renal stones and other chronic and mental illnesses not obvious on physical
inspection of the slaves and for which the window period is much shorter at 6 months.(46, 47)
Thus Plato allowed a longer period for the evolution of epilepsy compared to other diseases.
22
However, Hippocrates (c.460–377 BC) whose medicine had a particular focus on prognosis had
earlier separated epilepsy into the prognostic categories of childhood-onset and adult-onset
epilepsy: childhood onset epilepsy is initially fatal, but spontaneous remission occurs in children
as they mature, and adult onset epilepsy has better prognosis as it does not lead to death.
Hippocrates also thought that status epilepticus portends poor outcome and that early
treatment of epilepsy could lead to better prognosis.(47, 48)
In spite of the emergence of effective anti-epileptic drugs in the late 19th century and early 20th
century, accounts from Gowers(49) and Rodin(50) indicate there was much less optimism until
modern times regarding seizure outcome. Kwan and Sandler(51) have suggested this may be
due to much research into the prognosis of epilepsy being predominantly small-scale and
hospital-based retrospective studies which tend to result in bias as the success of follow up
often depends on a patient having poor seizure outcome.
SEIZURE OUTCOME IN UNTREATED POPULATIONS More recent studies of the prognosis of epilepsy have revealed an increasingly positive message,
buttressed by evidence of spontaneous remission of untreated epilepsy in some low and middle
income countries with large proportion of untreated patients.(51) The treatment gap for active
epilepsy exceeds 75% in low-income countries and 50% in middle-income countries whereas
high-income countries have treatment gaps of less than 10%.(52)
In a study that ascertained the duration of active epilepsy at diagnosis in Malawi,(53) there was
a steep progressive decline in the duration of active epilepsy at presentation. Much fewer
people had epilepsy for longer than 15 years compared to those with epilepsy for a shorter
duration at diagnosis. A survey in northern Ecuador showed that 31% of the total population
were spontaneously seizure free for at least 12 months in the period immediately before
assessment.(54) In rural China, about 40% of those in remission had never received anti-
epileptic drug therapy, (55) and in rural Bolivia, after a 10 year follow up, 40% of those who have
achieved 5 year remission had not taken anti-epileptic drugs for more than 1 year.(56)
While these studies are retrospective and rely on only clinical history to diagnose epilepsy with
case identification and classification that may not conform to international standards, the
explanation for the favourable prognosis of epilepsy from these studies cannot be entirely due
to poor case ascertainment. Life expectancy is lower in low and middle income countries, and
this fact may bias the result of the Malawi study. However, average life expectancy in China has
been comparable to that obtainable in the other high income countries in the past 10 years.(57)
The results are also in keeping with a study from Finland in which 42% of people with untreated
epilepsy achieved 2 years of remission by 10 years after onset.(58) An earlier study in Poland
also found about 30% of people who had not received any anti-epileptic drug therapy had
achieved remission.(59) These results may however be partly due to a selection bias as a study
23
in Vietnam found that the most common reason given by patients for never taking AEDs, or for
deciding to discontinue, was because their seizures were too infrequent to warrant the trouble
and costs of treatment, (60) a finding consistent with that from the Finnish study that patients
with milder forms of epilepsy were more inclined to reject anti-epileptic drug therapy compared
with those with more frequent seizures.(58)
Much has changed since the 1881 comment by Gowers(49) that “the spontaneous cessation of
the disease is an event too rare to be reasonably anticipated in any given case” although he
conceded to the more favourable prognosis in patients with newly diagnosed epilepsy treated
with bromide. Following the introduction of bromides in 1857, was among others, phenobarbital
(1912), phenytoin (1938), ethosuximide (1955), carbamazepine (1963), sodium valproate
(1967),(61, 62) and subsequently the newer generation of antiepileptic drugs.(62) Most people
with epilepsy in the high income countries receive antiepileptic drugs early in the course of their
illness, a fact that limits what is possible in the study of prognosis of untreated epilepsy.
SEIZURE OUTCOME IN TREATED POPULATIONS The natural history of epilepsy within populations of people that are treated with antiepileptic
drugs has been extensively studied. There are reports of 65 – 80% of patients entering long-term
remission on or off antiepileptic medication (21, 32, 63-70), of 45 – 55% maintaining remission
off antiepileptic medication (32, 71, 72) and 10 – 20% satisfying the definition of intractable
epilepsy (73-77). This thesis will explore the study characteristics that might explain the wide
discrepancies in the proportion of patients with the seizure outcomes in these studies. There is
also a group of patients who neither attain remission nor satisfy the definition of medical
intractability identifiable in some of these studies(72, 76, 78, 79), a group which Camfield et
al(80) named “something between.”
Most of the studies of prognosis in epilepsy in treated populations do not provide their results in
terms of seizure outcome by strategy of anti-epileptic drug treatment employed. Hence the
relationship between outcome and course of antiepileptic medication is not as widely known.
However, a hospital-based retrospective study by Kwan and Brodie(81) in Scotland shows that in
a patient group aged between 9 to 93 years, 47% of all patients treated were seizure-free for at
least 1 year at the time of last follow-up on the first AED, 13% on the second and only 4% for any
further AED regime: a total of 64% of the cohort was seizure free. Moreover, similar to the
report by Kwan and Brodie(81) were findings from a prospective hospital-based Dutch childhood
cohort (66) in which 46% achieved remission (seizure-free for at least 1 year at 5 years of follow-
up) on the first AED, 19% on the second and 9% on any further AED regime: a total of 74% of the
cohort was in remission. In the childhood Dutch cohort, 58% of the patients achieved remission
after the first antiepileptic drug failed compared to the Scottish cohort that comprised people
aged between 9 to 93 years, where 32% achieved remission after first AED failure.(66, 81) The
difference between the results of these studies may support what Hippocrates had suspected
and Gowers suggested; that prognosis may be better in childhood onset epilepsy. However, this
24
inference is limited by the study with fewer patients in remission being a retrospective study,
while in the study with better outcome, patients were prospectively identified.
In a meta-analysis (81-83) that included the Scottish study(81) involving 621 patients overall
with newly diagnosed epilepsy, Wiebe(84) showed that 48% of patients became seizure free
with the first AED and of those who failed the first, 27% responded to a second drug, and of
those who failed a second drug 12% responded to more than 2 combined AEDs. These results
relating to drug failure and number of antiepileptic drugs required to achieve remission
notwithstanding, the choice of specific drug used has not been shown to significantly influence
seizure outcome.(85-87)
No controlled trial has ever shown that any of the first line AEDs is better than the others in
terms of achieving remission, often leaving the physician to choose which drug, mostly single, to
use initially based on a case by case consideration of cost (of the drug and necessary follow-up
investigations), and potential for adverse effects.(88) The AED dose is then usually increased
gradually to a near toxic level before considering a second drug on account of not achieving
remission, allergic or other adverse reaction. The second drug is titrated to therapeutic levels as
the first drug is tapered and discontinued. Subsequent substitutions are done in similar fashion
before polytherapy is considered; an approach that effectively limits the additive adverse effects
of multiple antiepileptic drugs.(89)
The course of epilepsy in treated populations therefore has been shown to be good, although
not as fixed as these studies might suggest. There has been evidence of a remitting-relapsing
course for certain individuals who enter remission only to experience subsequent recurrence.
Patients with remitting only course achieve uninterrupted terminal remission, early when within
12 months of intake or commencement of therapy or late when after 12 months as defined by
Sillanpaa and Schmidt(25). There is a subset of patients running a remitting course who do not
have another seizure since intake into a cohort on or off medication, a subset described as
“smooth-sailing epilepsy” by Camfield et al (90)
In the Nova Scotia, Canada cohort, Camfield et al (71, 90) reported that 21% of 472 children
(excluding those with absence, myoclonic, or atonic seizures) had smooth-sailing epilepsy, and
an overall 48% had a remitting course. In the Dutch cohort,(66) 14% of 453 children had smooth
sailing epilepsy, while altogether 41% had a remitting course. The results are similar in the
Finnish cohort with 16% of the patients having smooth-sailing epilepsy and 48% overall with
remitting course.(25) Those with a remitting-relapsing course have their remission interrupted
by a relapse; some of them however go on to achieve terminal remission. Of the children who
experienced relapse in the Finnish cohort, 58% achieved terminal remission, 52% after 1 relapse
and only 6% after 2 episodes of relapse.(25)
At the other end of the spectrum are those whose seizures will become medically intractable,
another fluid category. Berg et al(73) defined intractability as “failure, for lack of seizure control,
25
of more than 2 first-line anti-epileptic drugs with an average of more than 1 seizure per month
for 18 months and more than 3 consecutive months seizure-free during that interval.” In their
cohort based in Connecticut, USA, 12% of children who had earlier satisfied the criteria for
intractability went on to attain remission.(73)
In the Dutch cohort, 6 out of 25 (24%) children who had satisfied a similar definition of
intractability at 2 years follow up, attained 1 year terminal remission at 5 years, and 18% of
those who had been seizure free for 1 year during follow up did not achieve 1 year terminal
remission at 5 years, and 2% had even become intractable.(66) In another childhood cohort, 4%
of patients who had experienced recurrent seizures for 2 years achieved remission with each
year of follow up.(91)
Therefore patients who have been seizure-free for a particular period of time at the point of
assessment during follow up may not be categorised as being in such a category at a different
point, as is the case for those satisfying the definition of medical intractability. The choice of end
point for analysis varies from study to study. Hence the pictures we have from different studies
are like snapshots, taken every 2 or so seconds, of for example the screen during a scene in a
motion picture. Most things might remain in their position, but certain things, no matter how
few are bound to change or be different – a shift, a change in the position of the mouth, a smile
becoming stilled in a frown.
This is the state of affairs in prognosis research in epilepsy. While research does not tell exactly
what the prognosis of epilepsy is – it will take a report on each patient in a cohort to achieve
that – we at least have snapshots from which we hope we might understand the specific
predictors of each state covered in the snapshots. Several prognostic factors identifiable at the
time of diagnosis have been shown to distinguish between those most and least likely to
experience a particular seizure outcome. They include symptomatic aetiology, history of status
epilepticus, multiple types of seizures, younger age of onset, having many seizures before
initiating medication, and failure to respond early to medication.(21, 32, 63-79, 81, 92-107)
1.2.2 WHEN TO COMMENCE OR DISCONTINUE MEDICATION This segment presents information from previous studies regarding making the decision about if
and when to commence anti-epileptic medication in patients with epileptic seizures.
Antiepileptic medication is readily availability in high income countries where much of the
studies of the prognosis of epilepsy studies have been conducted. Hence, there has been little
opportunity to study the course of untreated epilepsy. Ethical considerations have also limited
prognostic research that may help determine the proportion and characteristics of people with
epilepsy who enter spontaneous remission. However, it has been shown in clinical trials that
antiepileptic drug prophylaxis in patients with severe head injury(108, 109) and craniotomy(110)
only suppress immediate seizures, but not the ultimate development of epilepsy in those at risk.
26
The Italian FIRST (First Seizure Trial) study randomised 419 people with a first tonic-clonic
seizure into immediate and delayed (until after a recurrence i.e. diagnosis of epilepsy) AED
treatment groups without significant long term difference in probability of achieving
remission.(111) In the MESS Trial, 1443 people with single or infrequent seizures for whom
indication to commence treatment was not clear were randomly assigned to immediate or
delayed (until physician and patient agree AED is necessary) treatment. Immediate AED therapy
reduced the occurrence of seizures in the first 2 years, but did not affect long-term prognosis in
this group of patients.(98)
However, in the short term, next to the prognostic category of people with epilepsy who will
enter spontaneous remission is the group of those who will enter remission only with AEDs and
continue in remission even after the drugs are withdrawn. This category can only be determined
in studies designed to phase out the use of AEDs in patients in remission. Much of prognosis
research in epilepsy had been on determining the predictors of successful AED withdrawal. The
decision to discontinue AEDs may be more difficult than the decision to start the drugs.(112)
The benefit of seizure prevention in the short term does outweigh the risk of potential adverse
effects, the cost, stigma and inconvenience associated with AEDs.(113)
However, the question as to whether the benefits of AEDs still outweigh the drawbacks of AEDs
arises for patients in long-term remission. The possibility of the development of serious adverse
events is a strong argument for discontinuing AEDs, especially in children in remission who in
their formative years are on drugs that affect their cognition, and women in their child-bearing
years who are in remission and intend to have children as teratogenicity is a potential problem
with most of first line AEDs.(112) The discontinuation of AEDs is considered in the light of
implications of recurrence for patient safety, driving privileges, employment and liability.(113) It
is important to understand the predictors of recurrence clearly before embarking on AED
withdrawal.
With rate of relapse following AED discontinuation ranging from 12 to 63%(113) in a wide array
of studies with different designs, Berg and Shinnar(114) conducted a meta-analysis of 25 studies
that met strict inclusion criteria, and found that the risk of relapse was 25% (95%CI 21% to 30%)
after 1 year of initiating AED withdrawal, and 29% (95%CI 24% to 34%) after 2 years. The meta-
analysis also found that the rate of relapse was higher in adult-onset epilepsy compared to
childhood-onset epilepsy. The patients with remote symptomatic epilepsy and those with
abnormal EEG also had a higher rate of relapse. However, the meta-analysis by Berg and
Shinnar(114) addresses relapse after initiating withdrawal, and not after completion. Therefore,
it does not quite tell us the percentage of patients who remain in remission after completing the
AED withdrawal process. In the Nova Scotia Canada cohort 48% of the entire cohort was able to
discontinue AED therapy and remain in remission(71). This was in keeping with the British MRC
Antiepileptic Drug Withdrawal Study which showed that within the group of patients with at
27
least 2 years terminal remission randomised to slow discontinuation of AEDs, about 50% were
successfully discontinued(115).
Further analysis and follow up of the MRC study cohort showed that the rate of seizure
recurrence after AED withdrawal was same for all the drugs used in the cohort (phenobarbitone,
phenytoin, and sodium valproate) except cabamazepine for which the rate of recurrence is
lower(116) It was also found that AED discontinuation does not modify the long-term prognosis
of a person's epilepsy, although it increases the risk of seizure 2 years following discontinuation,
(117) a finding that further confirms the time dependent beneficial effects of antiepileptic drugs.
1.2.3 WHEN TO CONSIDER OTHER INTERVENTIONS This segment clarifies the definitions of uncontrolled seizures adopted for the purpose of this
thesis and presents the literature describing the justification for early consideration of non-
pharmacological interventions such as epilepsy surgery and ketogenic diet in people with
medically intractable seizures.
DEFINITIONS OF UNCONTROLLED SEIZURES Most studies refer to uncontrolled seizures using the term intractable or refractory, and some
other studies use both terms interchangeably.(72-79) The definition of intractability is usually a
variation on a “failure, for lack of seizure control, of more than 2 first-line anti-epileptic drugs
with an average of more than 1 seizure per month for 18 months (some 2 years) and no more
than 3 consecutive months seizure-free during that interval”(78) The variations on this definition
of intractability is adopted in this thesis as well as the use of the term “intractability”.
However, the fact that the term intractable is often used interchangeably with refractory is a
source of confusion. In this thesis, the term refractory is used to denote the category of
patients who had not achieved at least 1 year seizure-free period immediately before the last
assessment. This is in keeping with the definition from Semah et al(103), Kwan and Brodie(81),
Stephen et al(118) and Hui et al(97). The broader category defined as refractory therefore also
effectively includes within it patients who will fulfil the criteria for intractability.
SEIZURES BEGET SEIZURES The 1881 aphorism by Gowers that “seizures beget seizures”(49) may be wrong in much of
human epilepsy. However, there are certain patients for whom it does seem apposite. First in
this category is evidence from a study by Hauser and Lee(119) that in a first unprovoked seizure
cohort, individuals with low risk for seizure recurrence in the cohort (idiopathic/cryptogenic
aetiology) demonstrated a significant increase in risk for seizure recurrence with increasing
numbers of seizures. However, since the majority of these patients will ultimately achieve
remission and discontinue AED successfully, the authors argued with merit that there must be
competing forces that increase or decrease risk for seizure recurrence, of which the number of
previous seizures is only 1 of several contributors.
28
In a further analysis of the Nova Scotia, Canada cohort by Camfield et al(120) children with more
than 10 seizures (who were also more likely to have focal seizures associated with impaired
consciousness) were less likely to enter remission. Shorvon and Reynolds(121) had earlier shown
that the number of focal seizures associated with impaired consciousness occurring prior to
treatment was dramatically different: 20 (range 2-180) in those who achieved remission and 73
(range 20-960) in those whose seizures did not remit.(122) These 2 studies suggest that Gowers’
“seizures beget seizures” may also be true for patients having focal seizures associated with
impaired consciousness. Hence, considered alongside the argument for competing forces by
Hauser and Lee(119), Gowers may have been at least partly right in these specific groups of
patients.
The continued occurrence of seizures also changes the structure of the brain. Multani et al(123)
have shown that changes in neuronal ultrastructure are associated with relatively large numbers
of seizures over long periods of time. They used computerized 3-dimensional image analysis to
evaluate the features of neurons removed as part of surgery for mesial temporal lobe epilepsy
(TLE). More synaptic neuronal elements were lost and the 3-dimensional complexity of the
neurons was simplified in direct relation to the total lifetime number of seizures and the
distance of the brain tissue away from the seizure focus. Three groups that prospectively
assessed every 3 to 4 years the degree of cerebral atrophy with magnetic resonance imaging
(MRI) in patients with chronic(124) and temporal lobe epilepsy(125, 126) demonstrated
progressive cerebral atrophy in these patients but not in those with newly diagnosed epilepsy or
those whose seizure are controlled.
PSYCHOSOCIAL ASPECTS OF UNCONTROLLED SEIZURES The continued occurrence of seizures is undesirable for reasons beyond the possibility of the
seizures begetting further seizures. Numbered among other reasons why uncontrolled seizures
are undesirable is the psychosocial impact on the lives of people with epilepsy.
In a cross-sectional study (N=696) by Jacoby et al(30) of British adults, 44% of 168 people with
frequent seizures (1 or more seizures per month in the past 3 months) had Hospital Anxiety and
Depression Scale (HADS)(127) case defining score for anxiety, compared to 13% of 350 without
current seizures. The depression case defining score on the HADS was met by 21% of patients
with frequent seizures compared to 4% without current seizures. In the same study, 62% of
people with frequent seizures felt stigmatised compared to 25% of those without current
seizures. Also, fewer people with frequent seizures were married, employed or had a sense of
social support. The fact that this study is cross-sectional study and does not use any of the more
usual definitions of uncontrolled seizures limits inferences that could be made from these
results.
29
However, Thompson et al(128) in a case-control study using the Patient Health Questionnaire-
9,(129) to determine clinically significant depression in another study in adults, also showed that
clinically depressed people with epilepsy were significantly less likely than non-depressed ones
to be married or employed and more likely to report comorbid medical problems and active
seizures in the past 6 months. This study is also limited by being a cross-sectional analysis.
In a prospective cohort study by Camfield et al(130) conducted on the intellectually normal
children within the Nova Scotia cohort, the proportion with 8 different adverse psychosocial
outcomes were reported thus – 34% had experienced school failure, 34% used special education
facilities, there was mental health consultation in 22%, unemployment in 20%, social isolation in
27%, undesired pregnancy in 12%, 5% were on psychotropic medication and 2% had been
convicted. When the children were divided into those with bad social outcome (1 or more of
the 8 adverse factors) and good social outcome (none of the adverse factors), stepwise logistic
regression analysis showed that more than 21 seizures before treatment was an independent
predictor of bad social outcome. Sillanpaa et al (32, 69) also showed in prospective Finnish
cohort that childhood-onset epilepsy is significantly associated with increased risk for
considerable psychosocial, vocational, and cognitive dysfunction, persisting into adulthood.
Jokeit and Ebner(131) found that the duration of epilepsy predicted global cognitive impairment
in a cross-section of adults with intractable temporal lobe epilepsy, where after correcting for
educational level the duration of epilepsy remained a more important independent predictor of
cognitive dysfunction than age, age of onset, site of epilepsy, aetiology, use of AEDs, and
occurrence of generalized seizures. Nolan et al(132) has confirmed this finding in a prospective
study of childhood epilepsy in Sydney, Australia with earlier age of onset as covariate of longer
duration of epilepsy. The exact same covariates were found by Pavone et al(133) among
children with the more benign absence epilepsy. In the Sydney childhood epilepsy series, higher
frequency of seizures, generalized symptomatic epilepsies, and larger number of AEDs were also
independently associated with lower overall intelligence.
These outcomes may be due to the seizures themselves or to toxic effects of anti-epileptic
drugs. It has been conclusively proved in 2 placebo-controlled, randomized trials involving
healthy adults(134) and people with epilepsy(135) that antiepileptic drugs contribute to
cognitive dysfunction, the major effects being impaired attention, vigilance, and psychomotor
speed.(136) These effects increase with higher dosages and number of antiepileptic drugs(137)
and subjective health status is affected more negatively by adverse effects of antiepileptic
medication than seizure frequency.(138)
The cause of these psychosocial outcomes is likely to be a complex interplay of underlying
neurologic abnormalities, effects of recurrent seizures, AED toxicity, and psychosocial factors
which are more prevalent in people with intractable seizures.(134-138) Therefore if ongoing
seizures, or the psychosocial factors and long running high doses of AEDs which are in
themselves consequences of persistent seizures, are responsible for cognitive dysfunction and
30
dysfunctional adaptation to life, then earlier non-pharmacological interventions could reduce
these immediate and long-term outcomes.(139)
ECONOMIC IMPLICATIONS OF CONTINUED SEIZURES There is also an economic argument to be made for early non-pharmacological intervention.
Heaney and Begley(140) in a systematic review of economic evaluation of epilepsy management
identified 3 published studies of the cost effectiveness of epilepsy surgery.(141-143) The
incremental cost of surgery was estimated at US$16,000 to US$27,000 per QALY (Quality
Adjusted Life Years), and the cost per seizure-free patient to be about 15% of AED use over a
lifetime in patients with intractable seizures. Begley et al estimate that in the United States, 80%
of the cost of epilepsy is attributable to patients with medically intractable seizures, which form
less than 20% of the total epilepsy population.(73-77) This disproportionate cost of people with
medically intractable seizures is likely due to the cost of continued AED use at high doses to
control seizures.
EARLY EPILEPSY SURGERY Epilepsy surgery is any surgical procedure carried out to control seizures in epilepsy. It may be
neurosurgical in which the seizure focus is resected, isolated or simulated. It may also be surgery
carried out to implant stimulator device and electrodes for Vagus Nerve Stimulation (VNS)
subcutaneously in the chest and neck.
Neurosurgical Epilepsy Surgery and Surgically Remediable Epilepsies
The benefit of resective surgery especially has been demonstrated repeatedly.(139) Early
epilepsy surgery may be life-saving as the risk of mortality in people with intractable epilepsy is
up to 5 times higher than in the general population reversing to normal in those who become
seizure-free after epilepsy surgery.(37, 144, 145) The risk of SUDEP (sudden unexpected death in
epilepsy) is zero with seizure remission but up to 1 per 100 per year in patients with frequent
and severe tonic clonic seizures.(8) Patients who become seizure-free after surgery also had
better cognitive function, social and behavioural outcome (131, 146, 147) but the effect of an
older age at surgery(146, 148, 149) or longer duration of epilepsy persisted even after successful
surgery.(131)
Epilepsy surgery is not without its own risks. However, careful candidate patient selection is
crucial in preventing these risks. Helmsteader et al(147) posits that the dichotomy of surgical
outcome is that patients either achieve remission with cognitive stability “double winners” or
achieve neither “double losers,” a sad outcome observed mostly in only poorly selected
candidates - older patients, and patients without hippocampal sclerosis who undergo dominant
temporal lobe surgery.(150) Otherwise, neurological deficits such as hemiparesis, visual field
defects, dysphasia are often temporary.(151)
31
Prognosis studies that focus on treatment decision making for epilepsy surgery are conducted in
cohorts limited to specific seizure types, syndromes or by virtue of having intractable seizures or
epilepsy surgery. (152-162) Téllez-Zenteno et al(163) have just completed a meta-analysis to
investigate the surgical outcome in 697 patients with non-lesional epilepsy and 2860 patients
with lesional epilepsy. They found the odds of being seizure-free to be 2.5 (95%CI 2.1, 3.0) times
higher in patients with lesions identified on MRI or by histopathology. The outcome was similar
when they disaggregated studies by surgery type (temporal lobe and extra-temporal lobe
surgery) and by age (children and adults). A meta-analysis by Tonini et al(164) had earlier
suggested that having identifiable lesions – mesial TLE, tumors, and abnormal MRI – were the
strongest positive prognostic indicators of postoperative seizure remission.
Those patients with mesial temporal lobe epilepsy (TLE), localization-related epilepsies with
well-circumscribed lesions, and unilateral hemispheric disorders in infants and children have
been dubbed by Engel and Shewmon(165) as having “surgically remediable epilepsies.” Engel
(166)had initially put up mesial TLE as the prototypical surgically remediable epilepsy syndrome,
as it is the type mostly found among patients with medically intractable seizures who can expect
to become free of disabling seizures postoperatively.(151, 156, 167)
The concept of “surgically remediable epilepsies” mooted by Engel and Shewmon(165) was a
reaction to 2 emerging trends – the fact that new antiepileptic drugs might effectively render
the definition for medically intractable epilepsy irrelevant as the introduction of new drugs
makes it impractical to prove that any given patient’s seizures are intractable in spite of trying all
available medications in every possible combination.(168)Secondly, the time it takes from
diagnosis to surgery might have increased due to multiple drug trials in patients with little
chance of remission. (139, 168) The idea of “surgically remediable epilepsies” therefore
proposes to replace that of “medical intractability.” However while it is true that certain
epilepsies are recognisable as surgically remediable from an early stage, it is also true that there
are others who may not be so identifiable and for whom early surgery might be beneficial. In a
Hong Kong, China unselected childhood epilepsy cohort, Kwong et al(76)reported that only 57%,
of the children with medically intractable epilepsy had detectable abnormality on CT and/or
MRI. Ko et al(79) reported 55% in a Massachusetts childhood cohort and Berg et al(73) had 20%
from a cohort in Connecticut, USA.
Instead of replacing one with the other, there may be more benefit from having a standardised
definition of medical intractability which allows for context specific application especially for
“surgically remediable epilepsies” and which also accommodates the concept of remitting-
relapsing epilepsy course. For example, on the Scottish study by Kwan and Brodie(81) only in 4%
of the patients achieved remission on a third antiepileptic drug singly or in combination. This
suggests that the present commonly varied definition that limits intractability to more than 2
antiepileptic drugs singly or in combination may suffice for the pragmatic purpose of timely
determination of candidates for early epilepsy surgery.
32
Vagus Nerve Stimulation
In patients whose intractable seizures might not be neurosurgically remediable – especially
when the seizures have multiple cortical foci, when seizure focus is in the dominant or eloquent
hemisphere, or cannot be defined – or for other reasons not candidates for epilepsy surgery,
there is the possibility of Vagal Nerve Stimulation (VNS). VNS is conducted through a surgical
procedure that implants a stimulator in the chest, connected to bipolar electrodes wrapped
around the left vagus nerve in the neck. Saneto et al(169) reported a prospective study of 43
children with medically intractable seizures on VNS who were followed up for more than 12
months. They achieved an overall median seizure reduction rate of 55%. 37% had at least 90%
reduction. However, no patients are rendered completely seizure-free on the procedure.(8)
In a pioneering report of the analysis of follow up data of 440 participants from a series of open
label adult VNS trials (with the participants also on AEDs allowing for dosage adjustment as
required) between 1988 and 1995, Morris et al(170) demonstrated the efficacy and tolerability
of VNS in medically intractable seizures. Following implantation of a stimulator, there was more
than a 50% reduction in seizure frequency occurring in 36.8% of patients at 1 year, in 43.2% at 2
years, and in 42.7% at 3 years. The adverse effect profile –hoarseness and paraesthesias – was
tolerable with about 75% of participants still on therapy at 3 years follow up. Other adverse
effects may include dyspnoea, cough, left vocal cord paralysis and lower facial muscle paresis.(8)
KETOGENIC DIET Besides epilepsy surgery, the high fat low carbohydrate “ketogenic” diet (sometimes using
medium chain triglyceride to boost ketosis), in use since 1921, is another non-pharmacological
intervention to control intractable seizures. Its use is based on the theory that ketosis has
antiepileptic properties in the presence of low circulating glucose level.(171) Keene(172)
showed in a meta-analysis, that 15.6% of children with medically intractable seizures on
ketogenic diet were seizure-free and 33% had more than 50% reduction in seizures at 6 months
of follow up. However, patients often discontinue the diet and in a meta-analysis by Henderson
et al(173) the reasons for discontinuing the ketogenic diet were: less than 50% seizure reduction
(47%), diet restrictiveness (16%), and presence of incurrent illness or diet side effects (13%).
The only RCT comparing the ketogenic diet against no change in treatment reported on 3 month
seizure outcome in children with medically intractable seizures.(174) After 3 months, the mean
percentage of baseline seizures was significantly (75%) lower in the diet group than in the
controls. Also, 38% of children in the diet group had greater than 50% seizure reduction
compared with 6% control; 7% in the diet group had greater than 90% reduction in seizure
frequency compared with none among the controls.
Therefore, although with relatively poorer results, but wider applicability and less invasiveness,
VNS and the ketogenic diet remain viable and appropriate options in patients with non-
surgically remediable medically intractable seizures.
33
1.2.4 THE SEIZURE OUTCOME OF EPILEPSY The ultimate goal of the prognosis studies is to define specific risk groups based on prognosis
and to predict different categories of seizure outcome more “accurately or parsimoniously” in
order to provide tools for making informed treatment decision and patient counselling.(45)
Kwan and Sander(51) came up with an epidemiological synthesis of prognosis studies of the
natural history of treated and untreated epilepsy that suggests there are 3 prognostic groups
(see Figure 4). Group 1 (20–30%) are the patients with excellent prognosis. There is long term
remission after a period of seizures, with or without AEDs. The primary aim of AED therapy in
this group therefore is to suppress seizures until remission occurs, and patients are successfully
weaned off AEDs. For group 2 (20–30%) seizure remission occurs and is maintained only with
AEDs. In group 3 (30–40%) there is continuing seizures despite AEDs with some patients having
frequent enough to qualifying them as being medically intractable.
Figure 4 | Natural History of Newly Diagnosed Epilepsy Group 1 will enter “Spontaneous” remission (20-30%), 2 remains in remission only on AED treatment (20-30%) and 3 continues to have seizures (30-40%) [Adapted from Kwan and Sandler(51)] However, a further interpretation of available evidence and patterns of seizure outcome suggest
that this view could be further broken down to achieve a finer granularity in the prognostic
groups that people with newly diagnosed epilepsy could be predicted to belong at the point of,
or within 1 year of diagnosis. Group 1 is the group of patients who would enter remission with
or without AED therapy, those who would automatically achieve remission even if they did not
have access to AEDs.
Sub-Group 1A therefore will represent those with immediate spontaneous remission (ISR).
(Figure 5) They will not have a third seizure, i.e. upon starting AED or being recruited into the
cohort under consideration they enter terminal remission immediately, and will not relapse
after complete AED withdrawal. In a New York childhood first seizure cohort, Shinnar et al(24)
reported that out of 182 children who had a second seizure, 28% did not have a third seizure
(second recurrence) at 5 years follow up, meaning they could be effectively considered as
having entered into immediate remission.
PEOPLE WITH EPILEPSY
REMISSION
1.REMISSION OFF AED 2. REMISSION ONLY ON AED
NO REMISSION
3. NO REMISSION ON AED
34
Figure 5 | Natural History of Newly Diagnosed Epilepsy (with additional granularity) Sub-Group 1A will not have a second recurrence (third seizure) and will remain in remission after complete AED withdrawal. 1B enters remission, although not immediately and remain in remission after complete AED withdrawal). 2A achieves and stays in remission only on continued AED. 2B has 1 or more periods of relapse, but achieves remission ultimately and stays in remission only on continued AED. 3A has 2 different categories within it: i – those who have 1 or more periods of relapse, but do not achieve remission ultimately, ii – those who have never achieved remission, but who do not fulfil the criteria for medically intractable seizures. Sub-Group 3B medically intractable seizure which could also have within it the 2 categories of: i –those whose seizures will remit on non-pharmacological intervention, and ii – those who will never achieve remission no matter what is done.
Hauser et al(175) reported that 27% did not have a second recurrence after 4 years of follow up
in an all age cohort in Minnesota. Treatment did not influence the rate of second recurrence in
both studies, although for Hauser et al(175) the risk of additional seizures in those with 2
previous seizures is greater than the risk of adverse effects of antiepileptic drugs. Neither study
investigated the subset of these patients who are successfully taken off antiepileptic drugs.
Hence the proportion of patients within this Sub-Group may be less than they have reported.
Sub-Group 1b thus represents those who although they do not enter remission immediately,
ultimately do so, and who alongside those who enter immediate remission, will remain in
remission on antiepileptic drug withdrawal. There is however no study from which an estimate
of the proportion of patients who will be in this group could be derived. It does however fill an
important conceptual space within the range of temporal outcomes of seizures in newly
diagnosed epilepsy.
Notwithstanding their original eligibility into any of the Sub-Groups of Group 1, those who
cannot be successfully weaned off antiepileptic drugs belong in Group 2. It could also be
imagined that those within Sub-Group 2A are those whose epilepsy alongside the whole of
group 1 did run a remitting course only that they could not be taken off AED. Those in Sub-
Group 2B arrived in remission after a remitting-relapsing course. People in Sub-Group 2B made
up 20% of the Finnish cohort.(25)
Group 3 comprises those patients who will not enter terminal remission either because after a
remitting-relapsing course they do not achieve terminal remission (14% of Sillanpaa et al’s
Finnish childhood cohort) or because they continue to have seizures from diagnosis in spite of
AEDs i.e. they never experience remission. Within this group will be those (Sub-Group 3A) who
will continue to have seizures, not frequent or debilitating enough to be described as intractable
PEOPLE WITH EPILEPSY
REMISSION
1. REMISSION OFF AED
1A. NO 3RD SEIZURE 1B >2 SEIZURES
2. REMISSION ONLY ON AED
2A NO PRIOR RELAPSE 2B PRIOR RELAPSE(S)
NO REMISSION
3.NO REMISSION ON AED
3A RELAPSE OR NO RELAPSE
3B INTRACTABLE
35
and those (Sub-Group 3b) who will fulfil the definition of medically intractable seizures (10-20%)
(73-77).
Further subdivision of 3B would give us the 2 categories of those whose seizures will remit on
non-pharmacological intervention, and those who will never achieve remission no matter the
intervention. Sub-Group 3A will also have a sub-population of those who ran a remitting-
relapsing course within it. Therefore the whole of Sub-Group 2B and part of Sub-Group 3A is
made of patients running a remitting-relapsing course, which also indicates that the boundary
between the 2 Groups will be a potential area of cross-over activity, and certainly an area where
multivariate regression snapshots will capture different pictures in time.
One group that may exist but is not covered is of newly diagnosed patients who initially enter
immediate spontaneous remission and then go on to have a remitting-relapsing course, but are
eventually successfully weaned off AEDs. However, evidence from Sillanpaa et al(25) suggest
that it is unlikely that such individuals exist. They show that the number of years before entering
5-year remission is an independent predictor of relapse. Therefore there is a low probability that
a patient with immediate spontaneous remission will go on to have a remitting-relapsing course,
and if they do, it may be ill advised to wean them off AEDs.
The ideal prognosis study in unselected populations of people with epilepsy will be a prospective
cohort which is followed up long enough and in good detail to achieve this level of granularity in
making out different prognosis groups that will help inform treatment decisions and patient
counselling regarding issues of if and when to start antiepileptic drugs, if and when to
discontinue antiepileptic drugs and if and when to consider non-pharmacological intervention.
The ideal study will also use multivariable approach to analyse for prognostic factors that will
help guide these decisions, and also provide externally validated tools (models) to estimate
outcome probabilities based on combinations of these predictors.(23)
36
1.3 SYSTEMATIC REVIEWS OF PROGNOSIS STUDIES The third part of this introductory chapter presents an overview of the role of systematic review
in medical research with its application to observational studies in general and prognosis studies
in particular.
1.3.1 THE ROLE OF SYSTEMATIC REVIEWS In an exposition on the role of systematic review in health research, Chalmers(176) opened his
foreword to Systematic Reviews in Health Care: Meta-analysis in Context1 by quoting the
physicist Lord Rayleigh: “If, as is sometimes supposed, science consisted in nothing but the
laborious accumulation of facts, it would soon come to a standstill, crushed, as it were, under its
own weight… The work which deserves but I am afraid does not always receive the most credit
is that [of] introducing order and coherence in which not only facts are presented, but their
relation to old ones is pointed out.”
The introduction of order and coherence into a body of research is the central role of systematic
reviews. Cook et al(178) defined systematic review as “the application of scientific strategies
that limit bias by the systematic assembly, critical appraisal and synthesis of all relevant studies
on a specific topic” and meta-analysis as “a systematic review that employs statistical methods
to combine and summarize the results of several studies.” The need for this method of research
arose out of the fact that science itself is a cumulative process. The fast pace, and almost
exponential increase in the accumulation of scientific papers resulted in a tendency to lose
information in an abundant mass of facts, necessitating that the results of research be
synthesised.(179)
However, the traditional /narrative review, synonymous with expert opinion has been shown to
have a tendency to reflect a singular, often biased opinion that also may lag behind and even
contradict research evidence.(180-184). Murlow in 1987 was the first to highlight that the
review of clinical research was largely poor and unscientific.(182) Murlow reviewed a sample of
traditional/narrative clinical reviews published in 4 leading American medical journals, and
demonstrated that most fell short of the standards for systematic reviews. McAlister et al(181)
did a follow up 10 years later, and found that the situation had only improved marginally.
There is a metaphorical view of systematic review as an epidemiological study, one in which the
subjects are published and unpublished studies rather than humans. The data collection involves
taking a representative sample of a well-defined population (of studies) or identifying every
member of the population if an unbiased sample is impossible.(185) Systematic reviews start
like all epidemiological studies with a good question, followed by developing a protocol to
answer that question, collecting, synthesising and analysing the data to arrive at objective
conclusions.
1 The book "Systematic Reviews in Health Care: Meta-analysis in Context" has had a defining role in the emergence of
systematic review and meta-analysis as a central and important tool in health research. (177)
37
1.3.2 HISTORY OF RESEARCH SYNTHESIS IN MEDICINE The application of systematic reviews and meta-analysis lagged in medicine, behind the social
science research fields of education and psychology. In 1933, Peters(186)published a systematic
review of a series of experiments on the impact of moral instruction on modifying character “in
order to bring them together into a single form so that we may draw inferences from the whole
set.” It was a psychologist, Glass(187) who coined the term meta-analysis “to refer to the
statistical analysis of a large collection of analysis results from individual studies for the purpose
of integrating the findings.” In these roles systematic review and meta-analysis helped separate
the wheat from the chaff of available research, performing interpretative functions as well as
serving as a means of systematising existing accumulated knowledge.
Nevertheless, medicine had much earlier, albeit isolated attempts at systematic review. In 1753,
James Lind, a Scottish naval surgeon, who had to deal with a series of reports about scurvy came
up with a treatise containing “an inquiry into the nature, causes, and cure, of that disease
together with a critical and chronological view of what has been published on the subject." For
him, "it became requisite to exhibit a full and impartial view of what had hitherto been
published on the scurvy” and “to remove a great deal of rubbish."(179) Two 18th century
European journals [Medical and Philosophical Commentaries (Edinburgh) and Comentarii de
Rebus in Scientia Naturali et Medicina Gestis (Leipzig)] published critical appraisals of available
research evidence.(188)
However, the 20th century pioneer of research synthesis in medicine was Pearson.(179) In their
1904 review of a typhoid vaccine, Simpson and Pearson(189) extracted data and conducted a
meta-analysis of correlation coefficients in 2 groups from 11 selected studies of immunity and
mortality among British soldiers serving in various parts of the world. In another typhoid review
3 years later in 1907, Goldberger(190) analysed available data on bacteriuria in typhoid fever in
Washington DC, in strict adherence to yet-to-be-formulated criteria for systematic reviews and
meta-analysis.
The terms systematic review and meta-analysis have not always meant what they presently
mean. The idea of research synthesis has gone through periods of evolution over which these
terms and others including “research synthesis” have connoted different things. The term
research synthesis was and is sometimes still used extensively by the social scientists who led
the development of the science and practice of review in the second half of the 20th
century.(179) Although the term “systematic review” was used during the first half of the 20th
century, even before "research synthesis,"(191) it is uncertain whether it was in the same sense
of what it grew to mean during the second half of the century.(179)
However, the term systematic review slowly gained wider usage and the method became more
popular and recognised, reaching a tipping point in the late 1990s owing to the earlier use of the
term by Cochrane(192) in his foreword to a popular 1989 compilation of research syntheses in
38
maternal health,(193) the creation of the Cochrane Collaboration in the early 1990s, and the
promotion of “systematic review” in contradistinction to meta-analysis, in the first and second
editions of Systematic Reviews in Health Care: Meta-analysis in Context. (194, 195)
1.3.3 CRITICISMS OF RESEARCH SYNTHESIS IN MEDICINE Nonetheless, there have been epistemological challenges to the claims of the systematic review
process to objectivity.(196)This view concedes to systematic review its assertion of methodical
discipline in adhering to an explicit and auditable protocol. The contention is that this
“procedural objectivity” does not remove the subjectivity of the process(197) as search
results,(198) quality assessment,(199) data extraction(197) and review conclusions(200) all vary
with reviewers. The objectivity of the process is then, it was argued, only in its “disciplined
subjectivity.”(196)
However, on the other end of the spectrum, there is the challenge that over-adherence to
protocols has sometimes lead to “empty reviews”(201) in cases where no studies meet inclusion
criteria, with authors missing the opportunity to offer insight into the quality of studies so
excluded. The contention here is that “bad” research can yield “good” evidence.”(202)
The same subjectivity charge has also been levelled against meta-analysis,(203) especially in the
interpretation of funnel plots(204) and the use of quality scores.(203)The idea of quantitative
statistical synthesis (meta-analysis) has drawn strident negative responses.(176) Expressions like
"shmeta-analysis"(205) “an exercise in mega-silliness”(206) and “statistical alchemy for the 21st
century”(207) were a rare display of invective in the medical literature. However, the effort at
distinguishing systematic review from meta-analysis by the editors of Systematic Reviews in
Health Care: Meta-analysis in Context (194, 195) and others was “to prevent the former being
discredited by poor versions of the latter”(208). They set meta-analysis in proper context as an
optional addendum to the systematic review process only to be used with proper discretion.
1.3.4 SYSTEMATIC REVIEW OF OBSERVATIONAL STUDIES These criticisms especially of meta-analysis were more directed at reviews of observational
epidemiology, than of interventional studies. From only 4 identifiable in the 1985 publication
year(209) to over 400 published in 1996,(210) the use of meta-analysis in observational studies
is now about as common as in randomised controlled trials. In a random sample of meta-
analyses published available on MEDLINE search in 1999, Egger et al(211) found that 40% of the
meta-analyses were based on observational studies (mainly cohort and case-control) studies, a
result that is in keeping with an earlier analysis by the same group, of studies published and
available on MEDLINE search in 1995.(212)
While the randomised controlled trial (RCT) remains the gold standard in the evaluation of
intervention, aetiological and prognostic hypotheses can hardly be tested in randomised
experiments; it is often only by observational studies or laboratory experiments that these could
39
be investigated.(212) However, in meta-analysis, the assumption of random variation in the
results between studies, while it may be plausible in RCTs (as they are conducted in usually
homogenous groups of patients and within largely regulated settings) is not quite as plausible in
observational studies where estimates of association may deviate from true underlying
relationships beyond variation due to chance, as it is a design that is more susceptible to the
effects of confounding, and the influence of biases.(185, 212)
The most important threat to the validity of results from a well conducted cohort study is
confounding, while a plethora of biases may plague the case-control studies. Even in instances
where confounding factors have been allowed for in statistical analysis, residual confounding,
where the confounder is unknown, unsuspected or could be reliably or precisely assessed may
remain.(211-214) Therefore, the systematic review of observational studies is for these reasons
encouraged over meta-analysis which however remains particularly attractive as it results in
precise and definite risk estimates when the magnitude of the underlying risks are small or
when the results from individual studies disagree.(212)
However, the more specific application of systematic review and meta-analysis to prognostic
factor studies, while it shares the issues that plague that of observational studies in general, also
presents its own unique problems:(45, 211, 215) greater difficulty in identifying all relevant
studies compared to RCTs, the problem of publication bias (it is less likely that non-significant
results will be reported), and many of the studies are retrospective thereby increasing the
probability of bias.
There is the issue of inadequate and incomplete reporting of methods and outcome measures,
and the lack of recognised criteria for quality assessment. There is also wide variation and
inconsistency in study design, inclusion criteria into cohorts, methods of assessing prognostic
variables, approach to handling continuous variables, the choice of variables to adjust for,
methods of statistical analysis and adjustment for confounders, length of follow up, and the
presentation of results by different time points during follow up.(215)
1.3.5 SYSTEMATIC REVIEWS OF PROGNOSIS STUDIES IN EPILEPSY The conduct of prognosis studies in general has been frequently criticised within the medical
literature.(23, 39, 45, 216, 217) The contention has often been that in spite of its methodological
demands, prognostic studies are usually not protocol driven, small sized, widely heterogeneous
in characteristics of patients, choice and number of predictor variables, outcome and follow up
measures, and often prognostic analyses are conducted post-hoc in studies designed for quite
another purpose. However, the same issues are not restricted to prognosis research alone, but
have broader implications for observational studies generally.(218, 219)
The results of prognostic studies are therefore largely fraught with limitations, and with the
attendant uncertainty about the reliability of conclusions of their synthesis.(42, 217) This led to
40
the creation a new Cochrane Prognosis Methods Group in 2008 aimed at providing support and
a forum for discussion to facilitate and improve the quality of systematic reviews of prognosis
research.(220) Systematic review of prognosis studies with special focus on methodology is
therefore a burgeoning field of interest, and there have been efforts at ensuring the quality of
such reviews. Hayden et al(39) have developed a set of criteria, for quality assessment through a
meta-review of systematic reviews of prognosis studies. These criteria were adapted for use in
the quality appraisal of studies included in this review.
Hitherto, the systematic reviews in the literature of observational studies in epilepsy have only
focused on the incidence and prevalence of epilepsy generally(221) and with regional focus –
Asia(222), Latin America(223), Middle East and North Africa(224), Europe(225) and sub-Saharan
Africa(226). However, these studies, when they do, only make passing reference to the methods
and/or results of prognosis studies. Ross et al also systematically reviewed the literature, but on
management issues in epilepsy from 1980 to 1999.(227)The studies on prognosis were therefore
only partly considered and only in passing, without methodological rigour or consideration of
prognostic factors.
The studies on AED withdrawal have also been extensively reviewed(113) with meta-analysis of
prognostic factors conducted(114), as are of studies considering seizure outcome following
epilepsy surgery (163, 164)and the use of ketogenic diet.(173) However, in spite of the amount
of work that has gone into analysing prognosis in newly diagnosed epilepsy, there has been no
systematic review of these studies and neither has there been a meta-analysis of the prognostic
factors in patients with newly diagnosed epilepsy. However, beyond seizure outcome, no
systematic review has also considered other prognostic outcomes, especially of psychosocial
outcomes following newly diagnosed epilepsy, important as they are.
41
42
2 CHAPTER TWO: THE AIMS AND SIGNIFICANCE OF
THIS THESIS
This chapter presents the research question that the thesis aims to answer. The chapter also
discusses the central contribution of the thesis to the systematic review and meta-analysis of
prognosis studies in general and particularly to studies of prognosis in epilepsy.
The seizure outcome of epilepsy is varied. Epilepsy itself is a multi-aetiological and diverse
disorder. The first seizure brings with it the opening of a fresh chapter in the life of the patient,
with tough decisions to be made for the patient and for clinicians especially of when or whether
to start AEDs as they carry risks of acute idiosyncratic reactions, dose-related and chronic toxic
effects, and teratogenicity. For some of the patients newly diagnosed with epilepsy, the benefits
of treatment will far outweigh the risks associated with treatment, but there are those for
whom this risk to benefit ratio is more finely balanced.(98) There is also the group of patients
that may be candidates for non-pharmacological intervention as they will develop drug-resistant
intractable seizures.
There is thus the need for studies that identify independent predictors of seizure outcome.
These studies have appeared and accumulated in the medical literature. The studies are of
different quality and they employ varying methods, study population, outcome definition, and
duration of follow up. It is therefore useful to systematically identify all these studies, review
their methods and synthesise the results of those studies that have identified the independent
factors which consistently predict clinically relevant seizure outcomes in patients that have been
newly diagnosed with epilepsy early in the course of the disease.
There is a greater tendency for publication bias in observational research compared to
randomised trials(228) and particularly in prognosis studies, as it is more probable that studies
showing a strong, statistically significant prognostic ability will be published. This can lead to
invalid results and incorrect inferences.(229)There are also the biases associated with
observational studies. This systematic review will therefore thoroughly assess for quality of each
study in a reproducible manner, based on indices that reflect how prone the studies are to
different types of bias.
The systematic review method has been applied extensively to confirmatory prognostic factor
studies investigating contribution of one variable to an outcome. The methods for reviewing
exploratory studies that set out to identify the independent contribution of a range of variables
to an outcome have received less attention in the literature. However, studies of prognostic
factors in epilepsy are of the type that investigates more than 1 variable as there are no already
extensively established set of prognostic variables for seizure outcome in epilepsy. This type of
study presents its own unique issues and difficulties for the reviewer. Therefore this review also
43
offers an opportunity to explore these issues and to contribute to the understanding of how a
reviewer might confront these difficulties.
This thesis aims to answer the following questions:
1. What is the quality of prognosis studies that have attempted to identify independent
predictors of seizure outcome among unselected population patients with newly
diagnosed epilepsy?
2. What is the effect of study quality, especially potential risk for bias, on the results of the
prognosis studies?
3. Which factors have been consistently identified as independent predictors of seizure
outcome and which factors have been consistently identified as non-predictors?
4. Do satisfactory seizure outcome prediction models already exist?
The thesis also proposes to develop a classification scheme for prognostic factor studies in
newly diagnosed patients with epilepsy so that future studies can be fitted into the scheme and
their results interpreted more easily and readily within the context of previous studies of the
same category.
This thesis contributes to the present understanding of the methods of systematic review of
prognostic factor studies by developing and using a transparent and reproducible method of
quality appraisal and by showing how multiple prognostic variables might be handled in a
systematic review of studies that do not specifically investigate one particular prognostic
variable.
44
45
3 CHAPTER THREE: METHODS This chapter has within it a description of the systematic review process: eligibility criteria, the
search strategy, how each category of papers was excluded, and how studies from the same
cohort were handled. It also includes a discussion of the iterative process of developing the
quality appraisal and data extraction forms and how the results of each study were assessed for
quality i.e. tendency for bias. It also discusses the criteria for assessing the quality of externally
validated predictive models. This chapter also includes how predictor variables that are
consistent, independent predictors of particular outcomes were identified and categorised and
results compared in a meta-analysis where possible.
The guidelines for the design, performance and reporting for meta-analyses of observational
studies published by the MOOSE group(210) are followed in this systematic review and meta-
analysis. One of the defining characteristics of the systematic review – which itself is a
retrospective study subject to its own selection bias – is its fidelity to an a priori protocol. This
protocol is presented in Appendix I.
3.1 ELIGIBILITY CRITERIA For a systematic review of observational studies such as this, an overly strict set of inclusion and
exclusion criteria especially relating to study design may be unduly limiting. Having a strong
opinion about the best study design to answer the review questions may lead to selection bias
as it does not necessarily mean that these types of studies exist or that studies of better design
and higher quality are not available.(230) Therefore, published prospective and retrospective
studies – randomised controlled trials, cohort studies, nested case control studies and case-
control studies – of unselected (unrestricted) populations of people with newly diagnosed
epilepsy that assess the independent effect of predictors of seizure outcome using multivariate
regression analysis were sought for inclusion in the review, excluding as ineligible, case series
and cross-sectional studies.
Newly diagnosed epilepsy was defined as epilepsy within the first year for the purpose of this
review. Therefore only studies with predictor variables collected within the first year of onset,
diagnosis, presentation or treatment, depending on how intake is defined in each study.
Majority of patients in the studies will also have been followed up prospectively or
retrospectively for at least 1 year.
There is no consensus on sample size estimations for multivariable models. However the
standard rule of thumb is 10 or more events per variable (EPV) in the model to allow a robust
estimation of regression coefficients(231), a value supported by a simulation study,(232)
although a more recent study showed that it could be a bit lower(233). Since about 30% (21, 32,
51, 63-70) of patients with newly diagnosed epilepsy will not achieve seizure remission despite
46
continued antiepileptic drug therapy, a study of association of 2 to 3 predictor variables with
30% outcome rate requires at least 100 patients to achieve an EPV of 10 or more. Therefore to
be eligible, the studies also included must have at least 100 patients,(234) a figure also in
keeping with an inclusion criterion in a systematic review of prognosis studies by Hemingway
and Marmot.(235)
Only studies published in English language were included in the review. The decision to include
only English-language studies was made, considering the time, logistic and cost constraints of
translation, and also with the awareness that the influence of bias due to language of
publication is disputed, and its effect, when and where shown, has been little, and has been
considered only in reviews and meta-analyses of intervention studies.(236-239)
3.2 SEARCH STRATEGY There is no widely acknowledged optimal database and strategy for searching the literature for
prognostic studies.(45, 215) However, in a study to ascertain how many databases would be
necessary to ensure comprehensive coverage of observational studies in diabetes, Royle et
al,(240) found that no additional articles from English language journals were retrieved from any
database beyond MEDLINE and EMBASE.They found that MEDLINE alone retrieved about 94% of
all the articles retrieved from both databases and the overlap in journal coverage between
MEDLINE and EMBASE was 59%. In a much earlier study considering journal coverage, Smith et
al(241) in 1992 found the overlap between MEDLINE and EMBASE to be 34%. Therefore these 2
databases were searched to ensure comprehensive retrieval of prognosis studies in epilepsy for
potential inclusion in the review. With the wide overlap between MEDLINE and EMBASE, and
the fact that the indexing of search terms in EMBASE is more extensive with more synonyms,
compared to MEDLINE,(242) a more focused strategy was developed for searching EMBASE
using terms specific to study design and analysis.(243)
3.2.1 MEDLINE To develop search strategies that optimise the yield for prognostic studies in MEDLINE(244) and
EMBASE(245), Wilczynski et al (244, 245) used Ovid's search engine syntax for combination of
terms. The terms for best sensitivity (keeping specificity ≥50%) in MEDLINE (sensitivity 90.1%,
specificity 79.7%)2 was:
incidence (MeSH)
OR explode mortality
OR follow-up studies (MeSH)
OR prognos* (text word)
OR predict* (text word)
OR course (text word)
2 Sensitivity is the proportion of high quality articles that were retrieved while Specificity is the proportion of low
quality articles not retrieved
47
This result served as a guide in deciding the search strategy for the study. The process that
generated this result was not limited by specific clinical/disease terms. Therefore the
performance of the search may be improved by combining the search terms with content
specific terms using the Boolean ‘AND’. In a search from inception to March 2010, the terms
identified by Wilczynski et al(244) were combined with:
explode epilepsy (MeSH)
OR explode seizure (MeSH)
OR seizure disorder (text word)
3.2.2 EMBASE The reason why it is difficult to formulate search strategy for observational studies is that there
are many study designs and the terminology is not standardized.(246) Furlan et al(243)therefore
set out to identify terms in MEDLINE and EMBASE related to study design and analysis that
could help reverse identify relevant nonrandomized studies in 4 systematic reviews across 4
different clinical areas. They found that text word “multivariate” was 1 of 2 terms which limit
topic only searches in all 4 clinical areas. The text word “regression” (Cox regression and logistic
regression) was among others common to 2 of the 4 clinical areas.
These terms identified by Furlan et al(243) focusing on terms related to study design and
statistical analysis were used for the EMBASE search. The terms “multivariate” and “regression”
are common to the inclusion criteria for studies to be included in the review. The terms were
thus selected, in addition to “multivariable” to limit an “epilepsy” topic search. The search was
from inception to March 2010, and was not limited to EMBASE records alone; it also included
papers from MEDLINE present in EMBASE.
multivariate (text word)
OR multivariable
OR regression (text word)
AND
explode epilepsy (Emtree)
OR epilepsy (text word)
3.2.3 SCREENING THE RESULTS FOR ELIGIBLE PUBLICATIONS To reduce the likelihood of missing out any publication, the full list of titles and corresponding
abstracts from MEDLINE was independently screened for eligible papers by 2 reviewers (the
MPhil candidate and KB3, who has masters level training in epidemiology). Following the first
3 Ms Kathleen Bongiovanni, who had just completed her Master in Public Health (International), volunteered to join in
the first round of screening citations retrieved through MEDLINE on EndNote.
48
round of selection based on inclusion and exclusion criteria, the MPhil candidate returned to
EndNote library and conducted a second screening of titles and abstracts to ensure
completeness. However, for EMBASE, the first and second rounds of screening, selection, and
elimination was conducted by the candidate. The list of references of each of the potentially
eligible papers identified from MEDLINE and EMBASE was also manually searched by the MPhil
candidate.
3.3 DATA EXTRACTION In this review, individual studies (not publications) were the unit of analysis as there were
papers and cohorts that contributed more than 1 study to the review. For example, a cohort
might provide more than 1 study considering different categories of seizure outcome, or using
different methods of multivariate regression analysis to identify independent predictors of
seizure outcome.
The novel nature of the review precluded the use of any previously designed data extraction
form. Therefore a data extraction form was designed and initially piloted on 3 papers. The form
was redesigned and piloted on another 2 papers. Then finally, it was redesigned for a third time,
and piloted on 2 papers other than the previous 5.
The main issues that had to be addressed in the versions of the quality appraisal forms
concerned the areas of potential bias in prognosis studies as identified by Hayden et al(39)
which were successively edited to reflect the important quality issues in the studies included.
There was also a section on reporting characteristics of included studies in the third (final)
version of the form. The 3 versions are in Appendix II.
Data from each study were extracted by the candidate 3 times; twice into a data extraction form
(Appendix II), and during the third time, data were extracted and entered directly onto an Excel
spreadsheet database. The data extracted were compared with the previous extraction at each
succeeding stage and where there were discrepancies, clarification made by consulting the
paper(s).
The extracted data included authors and title of study, year of publication, journal, study design,
age range (including mean and standard deviation), study recruitment period, length of
reported follow up, epilepsy subtypes included or excluded, method of multivariate regression
analysis, investigated seizure outcome measures, and details of AED treatment and withdrawal,
and how decisions about AED and the choice of drug was made. The risk estimates of the
prognostic factors in the multivariable model, and corresponding confidence intervals were also
extracted for each study. All data were extracted from the published studies, including obtaining
related publications from the same cohort that authors have made reference to for further
49
details, especially of study population and methods. The authors however were not contacted
for further information.
3.4 QUALITY APPRAISAL There has been a wide range of scales, scores and checklists available to aid quality assessment
especially for systematic reviews of intervention studies.(247, 248) However, the use of quality
scoring in systematic review and meta-analyses is controversial as aggregate scores may
inappropriately conceal and conflate the elements that define quality and may thus not directly
be associated with quality.(203, 247-249) It has thus been suggested that key components of
design, rather than aggregate scores themselves, may be more important.(39, 249)
In keeping with the guidelines of quality appraisal in systematic reviews and meta-analysis, (39,
178, 195, 250) each eligible study was assessed for bias. Hayden et al(39) developed extensive
guidelines for assessing quality in prognosis studies on the basis of a framework of potential
biases. In a meta-review that identified quality measures used in systematic reviews of
prognosis studies, they pooled quality items into 6 areas of potential bias: study participation,
study attrition, prognostic factor measurement, confounding measurement and account,
outcome measurement and analysis. (Appendix IV)
These areas of potential bias are relevant to the purpose and question of the present systematic
review and were therefore adapted in the analysis of the methods of studies being reviewed.
However, the 2 areas of “prognostic factor measurement” and “confounding measurement and
account” were merged into 1 as this thesis is not a review of studies investigating the prognostic
value of one particular variable. These 5 areas of potential bias as identified by Hayden et al(39)
are discussed.
3.4.1 STUDY PARTICIPATION The area of study participation is concerned with adequacy of reporting and assessing how well
the population recruited into each cohort is representative of people with epilepsy within the
source population. This judgement is made from the authors’ description of study setting,
location and catchment area in the paper or a previous, referenced, publication. Therefore for
each cohort, answer (Yes, Partly, or No) was provided to the question “Is the source population
described?” The explicit description of inclusion/exclusion criteria and baseline characteristics of
the cohort was also assessed. This included the assessment of the definition of epilepsy in each
paper which according to International League against Epilepsy (ILAE) standard.(1, 2)
The combined information – from how well the cohort is representative of its source
population, inclusion and exclusion criteria, and description of the baseline characteristics of the
cohorts – is used to make a judgement as to whether recruitment of eligible individuals into the
cohort is adequate. Those studies that attempted to prospectively ascertain all the cases of
newly diagnosed epilepsy within a particular geographically defined population, with clearly
50
stated and appropriate inclusion and exclusion criteria are the ideal studies meeting the highest
quality standard within this area of assessment.
In retrospective case ascertainment, investigators go back over a period of time and identify
patients with recently diagnosed epilepsy for inclusion in the cohort. Prospective studies on the
other hand use on-going surveillance to identify newly diagnosed patients when they present
for medical attention, recruit them and subsequently follow them going forward. However,
studies with retrospective ascertainment of cases may have a prospective follow up component
to them, and some studies combined prospective and retrospective case ascertainment,
followed by prospective follow up.(44)
Retrospective case ascertainment is a potential source of bias in that identifying cases for
inclusion in study may depend on unfavourable outcome, as those who experience recurrence
will be easier to identify than those who remain seizure-free after a second or third seizure.(44)
It is also more likely that patients with prospective recruitment are more comprehensively
assessed for clinical variables and history of previous seizures at the point of diagnosis, thereby
avoiding under- or over-estimation of the number of seizures before clinical presentation, and
ensuring the general quality of the assessment potential predictors.(44)
The inclusion of single seizure patients, patients with a single episode of status epilepticus,
defining epilepsy by a third seizure, excluding patients with poor prognosis or those with
excellent prognosis within epilepsy cohorts reported and analysed in the studies are also
potential sources of bias. These deviations from standard inclusion and exclusion criteria were
highlighted where applicable.
3.4.2 STUDY ATTRITION This area of potential bias considers what happens after patients are recruited into the cohort
i.e. the proportion of the cohort completing the study at the end of follow up or the proportion
included in the analysis. The question to answer here is: are those lost during follow up similar
to those who completed the study? To make this judgement however, it is important to know if
there is an adequate proportion of cohort in the study analysis (76 – 100% Yes, 65 – 75% Partly,
<65 No, and when it is not stated in the report – Unsure).
This is considered, together with the reasons for loss to follow up and the characteristics of the
patients lost to make the judgment as to whether there are important differences between key
characteristics and outcome in participants who completed the study and those who did not
that may potentially bias the result of the study. For cohorts with >90% completeness of follow
up, without evidence of systematic pattern to loss in follow up, it was assumed that patients lost
to follow up are similar to those who completed the study.
51
However, not all the studies were designed to have the whole cohort included in analysis. In
some studies, a case-control analysis of the cohort was conducted instead. The comparison
groups in the classical cohort study of prognosis are people with a risk factor and all the other
members of the cohort without the risk factor. The relative proportion of individuals with an
outcome of interest in each risk factor group is then considered. The outcome groups – cases
and controls – are made up of the entire cohort.(28)
A variation on this is the nested case-control study in which the whole cohort is not included in
the analysis: the proportion of people with the risk factor under investigation is compared
between cases and matched controls or a random sample (sub-cohort) of those without the
outcome albeit from the same cohort.(251)This design shares all the advantages of the cohort
over case-control study design, especially when it is a prospective cohort: collecting risk factor
data before outcome, the time sequence of cause and effect, providing an unbiased estimate of
relative risk,(252-255) and the fact that cases and controls are from the same cohort,
notwithstanding the argument by Wacholder et al(256) that at least in theory, every case-
control study takes place within a “cohort” or population, albeit difficult to define.
However, the variant of the nested case control study design available to be included in this
systematic review was one in which the cases and control groups were defined by outcome, for
example patients whose seizure is in remission were controls, while for example those with
medically intractable seizures were treated as cases, thereby leaving out of the analysis patients
who do not satisfy both definitions.(75, 76, 78, 79, 94) This group of studies will subsequently be
referred to as “nested case-control” studies although technically, they are not altogether similar
in design to studies that bear that name in the literature.(251, 257, 258) In the conventional
nested case-control study, cases are identified in a defined cohort. For each case, a specified
number of matched controls are selected from among those in the cohort without the case
defining outcome.
However, especially in prognostic studies of newly diagnosed epilepsy, these studies do have
the additional advantage of comparing groups that are more homogenous, compared to studies
that involve the whole cohort. Within a cohort, there is a continuous range of severity of seizure
outcome. An assumption inherent in logistic and Cox regression is that the outcome being
considered in the analysis must be discrete. Therefore, by creating a dichotomy of outcomes –
remission and intractability – and excluding from analysis patients that form a continuum
between these 2 groups, nested case control studies achieve a better delineation of binary
outcome categories necessary for logistic and Cox regression analyses.
The judgment on attrition in these studies was also made with full understanding of their
peculiarity in selecting, based on set criteria in each study, the cases and controls from a defined
cohort. For these studies, if there was no evidence of systematic pattern to loss in follow up
beyond the choice of cases and controls, it was also assumed that patients lost to follow up
were similar to those who completed the study.
52
3.4.3 PROGNOSTIC FACTOR MEASUREMENT The definition of each prognostic factor has to be clear and specific to avoid bias. The question
was therefore asked: are the prognostic variables clearly defined? Answers were reported thus:
Yes, if all the prognostic factors were clearly defined or described, Partly, if not all, and No if
none of the prognostic factors were clearly defined or described.
There is also the probability that misclassification bias is introduced by virtue of the method by
which prognostic factors are assessed. The second question was if the methods of assessing the
prognostic factors were reliable enough to limit misclassification bias. In this case, it was
reported as Yes if there was a single site with single assessor, or a fixed protocol, or if there was
explicit information about training of assessors for multiple sites or if ILAE standards and
definitions (1, 2, 9, 11, 18) were rigorously followed even if there were multiple assessors and
there was no information on training assessors for multiple sites. In the absence of these 3
conditions, the answer was Unsure.
How adequate was the proportion of patients with complete data and how missing data were
handled in each study formed the third question in this area. Missing data could be in 3 forms:
missing completely at random (MCAR) when people with missing data are a random subset of
the complete sample of the cohort, for example the result of some patients’ brain computed
tomography (CT) scan is lost accidentally. The fact that data are missing is completely
explainable in terms of a random event as the probability that an observation is missing is not
related to the patients’ characteristics.(259, 260)
When missing data are MCAR, patients with complete data are also a random sample from the
cohort. Therefore a technique like simply deleting an entire observation if it is missing on any
item used in the analyses (listwise deletion) is appropriate as the only loss is statistical power,
and there is no apparent potential for bias in the estimates.(260) This is also true when data are
missing at random (MAR) in which case the probability that an observation is missing depends
on information for that subject that is present, i.e. the reason for missing data is based on other
observed patient characteristics,(259) for example if the number of seizures before treatment
cannot be ascertained in a random sample of patients with cognitive impairment, it could be
said to be missing at random (MAR), and could also be ignorable just like for MCAR.
However, it is not ignorable when data are missing not at random (MNAR). Here, the probability
that an observation is missing depends on the value of the observation itself,(259) like when the
reason that the number of seizures before AED cannot be ascertained is because the frequency
of pre-treatment seizures is high, and the patient or carer cannot recollect. In this case, there
would be a need to resort to the more complex imputation techniques to handle missing
data.(259, 260)
53
Therefore the question about if the proportion of patients with complete data was adequate
and how missing data were handled would have been reported in the following way: Yes when
data were complete, or when less than 20% of a prognostic factor data were MCAR (missing
completely at random) or MAR (missing at random) and listwise deletion or imputation was
used, or when less than 10% were MNAR with listwise deletion or imputation used. Partly when
MCAR and MAR was 20-30% with appropriate imputation or when MNAR was 10-20% with the
use of appropriate imputation. No when MCAR, MAR was more than 30% of a prognostic factor
or when MNAR (missing not at random) was greater than 20%. When these issues were not
addressed or reported, the answer was Unsure.
However, most authors did not report on these issues at all. Some authors included data
completeness as an inclusion criterion, making it impossible to assess if the study population is
representative.
How continuous variables are handled is also important in multivariate analysis. The choice is
between keeping them continuous or creating categories which may or may not be data
dependent. Categorizing patients into high- and low-risk groups based on an “optimal cut-point”
that maximises statistical significance or the non-data dependent division through the median
into 2 equal groups without a priori justification often leads to bias, loss of potentially important
quantitative information, loss of power and residual confounding.(261)
Therefore the question about how the continuous variables were handled was answered Yes for
papers that left all the continuous prognostic factors as continuous variables. However it was
Partly for those that kept some variables continuous and others categorised in a non-data
dependent and/or data dependent manner. It was also Partly for those that used only non-data
dependent method of categorising continuous data and No for those that used only data
dependent method or in combination with non-data dependent method.
3.4.4 OUTCOME MEASUREMENT It is also important that there is a clear operational definition of the seizure outcome being
considered in the study, with a long enough time frame to limit misclassification bias as seizure
outcome in newly diagnosed epilepsy has a temporal pattern, and when patients are followed
up longer it is more likely that they are classified correctly in their outcome category.(25) The
question about the clear definition of outcome measure is reported as Yes, Partly or No. It was
Yes if the outcome was clearly defined in relation to length of follow up and seizure-free period
or period of continued seizures. It was No if either the outcome was not clearly defined and if
the time dimension was not explicitly stated. It was Partly if the outcome was clearly defined but
the related time dimension was not stated explicitly.
The issue of being able to limit misclassification bias was assessed in a manner similar to that of
prognostic factor measurement. The method of assessment was considered sufficient to limit
54
misclassification bias if follow up for the majority of the cohort was for at least 1 year and it was
a single site with single assessor. The answer was Partly if there were 2 to 3 assessors in a site or
if there was 1 assessor in 2 to 3 sites. Multiple sites or multiple assessors without information on
training or standardization of outcome measure method was answered as a No.
For the third question in this area, the issue was about measuring the outcome and prognostic
factors blind to each other. It was assumed that in cohorts where cases were prospectively
identified, whether it was stated or not, the outcome and prognostic factors were assessed
blind to each other, and for retrospective case ascertainment, the answer is Unsure, unless
explicitly stated in the affirmative.
3.4.5 MULTIVARIATE REGRESSION ANALYSIS Multivariate regression analysis is the most commonly used method to examine and adjust for
variables in prognostic factor studies.(45) However, it is also commonly abused, especially as a
black box that simply produces risk estimates(262) without important details of the process and
sometimes, details of the results of analysis . The risk estimates tend to vary with the choice of
the mathematical model, coding of variables, and method of selecting variables.(263) It is
therefore important that information on the test of assumptions of model building be provided
such as a test for colinearity and interactions among variables.(263) It is also important to be
able to determine how sufficient the reported data are to assess the validity and reliability of
the results.
Therefore, the first question here was: how detailed were the results presented? If the risk
estimate or the regression coefficient was presented and the standard error or 95% confidence
interval of regression coefficients of risk estimate was presented, the report was Yes. However if
the result from only 1 of these 2 was presented, it was reported as Partly, and if neither, the
report was No.
The next question concerns whether the method of selecting variables to be included in the final
model was described. The choice of variables to retain in a regression analysis could be made by
forward selection (starting with a priori recognised predictor variables) or backward elimination
(starting with all available predictors, and could be all variables used in univariate analysis, or
only variables significant at (P < 0.05) in univariate analysis) or using the computer automatic
selection stepwise procedure (forward or backward), with different criteria for inclusion or
exclusion of a variable in the model. Important variables from clinical experience or previous
studies may be forced into the model as well as significance level fixed for prognostic factors to
retain or eliminate, in a forward or backward stepwise manner, manually or in automated
algorithms. Did the authors specify how predictor variables were selected? Was there a
preliminary screen based on the univariate association? Was some form of stepwise analysis
used, and if not, was colinearity between variables assessed? For this item, the answer is Yes if it
was stated that there was a stepwise selection of variables, and No if it was not so stated.
55
The risk estimates may however be unreliable if outcome events are too few relative to the
number of independent variables, leading to a situation where regression coefficients for
individual prognostic factors may represent spurious associations, or the effects estimated with
low precision.(231, 263) Therefore the rule of thumb, of having at least 10 events per variable
(EPV) is a liberal way of assessing if or not there is over-fitting of the data. This is done by
computing the ratio of the number of predictors in the model and number of patients with less
common outcome.(231)However it has been shown by Vittinghoff et al(233) that the rule of 10
could be relaxed a little.
Other issues related to statistical analysis that were also assessed include if the authors
attempted to justify study sample size, the number of patients lost to follow up and how loss to
follow up was treated in analysis.(263)
3.4.6 SUMMARY OF POTENTIAL FOR BIAS IN STUDIES
To assess the quality of each study, 1 summary measure of judgement about each of the 5
potential areas of bias (study participation, study attrition, prognostic factor measurement,
outcome measurement and multivariate analysis) was constructed for each of the individual
studies included.
The summary assessment for quality of study participation was the report of how adequately
the eligible individuals were recruitment of into each cohort, which is the answer to the third
question in the area of study participation. However, for study attrition, cohorts with >90%
completeness of follow up and nested case control studies, without evidence of systematic
pattern to loss in follow up, were assessed as adequate. Otherwise, studies were assessed based
on their report of how similar those lost to follow up are to those who are included in the
analysis. For studies with minimum follow up > 20 years, the rules were relaxed to allow for
more than 10% attrition rate.
To be assessed as adequate, methods that were assessed as sufficient to limit misclassification
bias must be used for prognostic factor measurement. The factors would at least be “partly”
defined and continuous variables at least “partly” kept continuous or have adequate proportion
of patients with complete data. The outcome must first be clearly defined, and the method of
measure must also be sufficient to limit misclassification bias, or measured blind to prognostic
factors and vice versa. Multivariate analysis was considered to be adequate only if the EPV was
at least 9.
3.5 SYNTHESIS OF RESULTS The consistently identified independent predictor variables were considered for synthesis. This
does not involve the statistical pooling of risk estimates; only the statistical conversion of risk
estimates into formats that would facilitate comparison in order to ascertain the consistency of
56
exposure-outcome relationship across different studies, and to make a decision on the least
biased risk estimate(s) for each outcome.(264) Variables that were found to be an independent
predictor in more than 1 study from different cohorts were considered to be consistent
predictors. On the other hand, the variables that were examined, but not retained in the model
in more than 1 study from different cohorts were considered as consistent non-predictors.
Hazard (or rate) ratio (HR) was assumed to be a reasonable approximation of the relative risk
(RR), as they are both simple quotients of “exposed” and “unexposed” quantities; HR a ratio of
rates and RR a ratio of probabilities.(265) However, the similarity or disparity of these measures
depends on the influence of the length of follow-up, which may impact on the average rate of
the occurrence of the seizure outcome of interest, and the risk of the “exposed” group relative
to the referent group.(265)
When association is weak and events are uncommon, relative risk, hazard rate ratio, and odds
ratio (OR) are good numerical approximations of one another; their disparity increases as events
increase and as risk departs from unity. However, the odds ratio is more subject to the influence
of these factors (265-267) as the hazard ratio reasonably approximates relative risk under a
much wider range of circumstances. Therefore, in studies where odds ratio was the risk
estimate, the Zhang-Yu equation(266) was used to shrink the odds ratio towards unity (in the
direction of relative risk), when event rate exceeds 10%, and/or odds ratio is greater than 2.5 or
less than 0.5.(265-267)
The Zhang-Yu equation(266) is such that in a cohort study, if P0 is the incidence of the outcome
of interest in the non-exposed group and P1 is the incidence in the exposed group:
Odds Ratio (OR) = (P1/1−P1)/ (P0/1−P0)
Therefore (P1/P0) = OR/ *(1 − P0) + (P0×OR)]
However, since relative risk (RR) = P1/P0
The corrected RR = OR/ [(1 − P0) + (P0 × OR)] (Zhang-Yu Equation)
Zhang and Yu(266) proposed the formula to estimate the risk ratio from the odds ratio in cohort
and cross-sectional studies with univariate and multivariable analyses. They validated the
formula with a simulation incorporating 2 confounding variables and it was shown that the
relative risk estimates were close approximations to the true relative risk. The equation was
recommended for use, and has been used in several meta-analyses.(265, 268-271)
The equation shows that as P0 approaches zero, the denominator approaches 1 and RR
approaches OR. This situation in which OR approximates RR obtains when the outcome is rare.
However, as P0 approaches 1, RR approaches 1 regardless of the value of OR, which shows the
57
large differences that occur between RR and OR when the outcome is common. When OR
equals 1, then RR also equals 1, regardless of the value of P0. For all P0 greater than zero and less
than 1, RR is less than OR when OR exceeds 1, and RR is greater than OR when OR is less than 1.
Therefore, as expected, the estimated relative risk is always closer to unity than the odds ratio.
(272)
3.6 EXTERNALLY VALIDATED PROGNOSTIC MODELS For studies with prognostic models that were externally validated, further quality appraisal will
be conducted with particular reference to the models. Laupacis et al(273) suggested additional
criteria specific to prognostic models in their 1997 paper which was an addition to previous
criteria suggested by Wasson et al (274) in 1985, many of which were again identified in the
more recent work Hayden et al(39). The factors peculiar to prognostic models from Laupacis et
al(273) that were not already discussed as one of the potential areas of bias from the study by
Hayden et al(39) are the sensibility of the model and its performance on internal and external
validation and in clinical practice. The models were assessed for their performance in internal
and external validation using the 2 parameters of calibration and discrimination.(23, 27, 275)
3.6.1 CALIBRATION AND DISCRIMINATION The performance of binary outcome models is usually assessed in terms of calibration and
discrimination. The calibration of a model is its ability to have predicted probabilities that agree
with the observed proportions of events over the whole range of probabilities. Calibration could
be investigated by plotting the observed proportions of events against the predicted risks for
groups defined by ranges of individual predicted risks, usually in 10 groups. Ideally, the plot
shows a 45 degree line in internal validation i.e. the slope is 1.(23) There is however a loss in the
calibration when the model is externally validated in another population. The statistical tests
sometimes used to assess calibration include Hosmer-Lemeshow test(276) and the calibration
component of the Brier score.(277)
The 2 probabilities that are often used to express the performance of binary tests were used to
assess the accuracy of the models as reported in internal and external validation studies: 1.)
Sensitivity or true positive rate (TPR) which is the proportion of actual positives [patients with
outcome, i.e. true positives (TP) and false negatives (FN)] that the model correctly identifies as
true positive (TP) (i.e. the percentage of newly diagnosed epilepsy patients that will enter
remission who are correctly identified by the model); 2.) Specificity or true negative rate (TNR)
which is the proportion of actual negatives [patients without outcome, i.e. true negatives (TN)
and false positives (FP)] which are correctly identified as true negative (TN) (i.e. the percentage
of newly diagnosed epilepsy patients that will not enter remission who are identified by the
model).(278)
However, 2 more clinically important probabilities were also used in assessing the accuracy of
the models. Positive Predictive Value (PPV) refers to the percentage of the true positive (TP)
58
among all those that the model correctly (TP) or incorrectly (FP) identified as positive (i.e. PPV =
TP/TP+FP); Negative Predictive Value (NPV) is the percentage of true negatives (TN) among all
those that the model correctly (TN) or incorrectly (FN) identifies as negative (i.e. NPV =
TN/TN+FN). The summary of the accuracy of the model is the proportion of patients within the
cohort with correct prediction, either positive (P) or negative (N) (i.e. TP+TN/P+N) The positive
and negative predictive values, unlike sensitivity and specificity change with pre-test probability
i.e. with the proportion of patients with the outcome within the cohort.(279) Therefore the PPV
and NPV measure in external validation reflect the accuracy of the model in a different cohort,
and a possible difference in pre-test probability.(279) PPV and NPV were computed from
available data when they were not explicitly provided.
However, before these probabilities are determined for a particular model, the ROC (receiver
operating characteristic) curve of the model in the development sample is plotted. The ROC
curve is a plot of the true positive rate (or sensitivity) versus the false negative rate (or
1−specificity) as the cut-off point that assigns a higher probability of outcome in the model is
progressively varied. In effect, it is a comparison of 2 operating characteristics (true positive rate
and false positive rate) as the discrimination criterion changes.(278)
The area under ROC (receiver operating characteristic) curve (AUC) is used to assess the ability
of a model to discriminate between individuals with and without the outcome being predicted.
The AUC represents the chance that given 2 patients, one who will develop an outcome and the
other who will not, the model will assign a higher probability of having the outcome to the
patient who will develop, for example, the seizure outcome of interest.(23, 275) The AUC for a
prognostic model is typically between about 0.60 and 0.85 (the values are higher in diagnostic
settings) (275) and it is usually deemed good if >0.70 and poor when <0.70 for prognostic
models. (107, 280)
The other purpose of the ROC curve is that it allows for the detection of an optimal cut-off
point, at which the combination of true and false positive rates is best for the purpose of the
model. Therefore values above this point will indicate a higher probability of the indicated
outcome, and lower values indicate lower probability of outcome. Ideally, this point is chosen to
assess the accuracy of the model in predicting outcome in the development cohort and in the
external validation, and the probabilities that assess accuracy at the cut-off point (sensitivity,
specificity, positive predictive values, negative predictive value and the proportion of patients in
each cohort with correct prediction) are computed.(107)
3.6.2 INTERNAL VALIDATION Although not explicitly stated and only implied by Laupacis et al,(273) internal validation is
important in assessing the performance of the model. However, internal validation does not
provide information about the model’s performance in another population. A model is internally
validated by splitting the dataset randomly into 2 parts. The model is developed using the first
59
part and its predictive accuracy is assessed on the second portion. If the available data are
limited, the model can be developed on the whole dataset and techniques of data re-use, such
as cross validation and bootstrapping, applied to assess performance.(27) In this review, the
results of the internal validation of the models that were subsequently subjected to external
validation were assessed for calibration and discrimination.
3.6.3 EXTERNAL VALIDATION It is important to validate a model intended for clinical use in a group of patients different from
the group in which it was derived, and preferably with different clinicians. This examines the
generalisability of the model.(27, 231, 273)This could be external in place and/or external in
time. When validation is external only in time, it is otherwise called temporal validation. In this
case, the performance of a model is prospectively evaluated on subsequent patients from the
same centre.(27, 281) However, a higher hierarchy of validation is when it is external both in
time and place,(281)as it is the only form of validation that could confirm the wider
generalisability of the model. The results of the external validation were assessed for calibration
and discrimination.
3.6.4 SENSIBILITY The evaluation of sensibility was based on judgment rather than statistical methods:
Is the model clinically sensible? Clinicians should think that the items in the model are clinically
sensible and that no important items are missing. It is difficult to determine which factors are
important in the prognosis of seizure outcomes in epilepsy, but this was determined
retrospectively following this systematic review. It was then subsequently documented whether
these variables were included in the model(234)
Is the model easy to use? This includes consideration of factors like time needed to apply the
model, and how simple it is to use. Models that require extensive calculations may be less likely
to be used than models with simpler scoring schemes.(216) Laupacis et al(273) are of the
opinion that prognostic models that recommend a course of action are more likely to be used
compared with those that simply describe the probability of an outcome. Therefore, it was also
assessed if the model was only reported as outcome probability or the authors also recommend
accompanying course of action. These 3 indices of sensibility were assessed and reported.
3.6.5 EFFECTS OF USE ON CLINICAL PRACTICE This refers to the prospective evaluation of the effect on clinical practice of using the prognostic
model in a patient population other than the one in which it was developed and validated to
show if physicians and patients are willing to use the model and how its use affects patient
behaviour and clinical outcomes. This is best done in randomised trials.(282) These criteria apply
to models that could be used to predict seizure outcomes in epilepsy, and were therefore used
in the analysis of methods of the studies with externally validated prognostic models.
60
61
4 CHAPTER FOUR: RESULTS
This chapter presents the result of the systematic review. The chapter starts by reporting the
results of the process of identifying eligible publications, followed by a description of included
studies, how they relate to the cohorts from which they are derived and the publications that
report them, and an account of the reporting characteristics of each of the eligible publications.
There is a detailed report of the quality appraisal of studies included from each publication. The
studies were subsequently disaggregated according to the seizure outcome predicted, and each
category of study is further evaluated and appraised with a view to: 1) Understanding and
explaining similarities and differences in results in relation to each study’s peculiar
characteristics and potential for bias, 2) Identifying consistent independent predictors and non-
predictors within each outcome category, and 3) Assessing the quality and performance of
externally validated prognostic models within each outcome category.
4.1 IDENTIFYING ELIGIBLE PUBLICATIONS When the search terms were run through MEDLINE, (from inception to March 2010) 14,967
citations were identified and exported to EndNote citation management software.
Following the first round of screening, 144 potentially eligible papers were identified after
eliminating duplications, review articles, editorials and commentaries, and papers with non-
seizure outcome. Papers with the study population restricted to a particular epilepsy subtype
(e.g. partial epilepsy or mesial temporal lobe epilepsy) or patient population (e.g. pre-operative
or post-operative epilepsy surgery candidates) were also eliminated. There were also papers in
which epilepsy or seizure was studied as an outcome of or in relation to another pathological
process (e.g. head trauma or autism) which were also eliminated. Reports of first seizure studies
without follow up to the second seizure and beyond, and studies of AED withdrawal were also
eliminated as were non-English language articles. (Figure 6)
The second round of screening yielded an additional 5 potentially eligible papers. Then the full
text of all 149 papers was assessed and 44 studies without relevant outcome, mostly non-
seizure outcome were eliminated, as were 66 prospective or retrospective cohort and 4 case
control studies without multivariate regression analysis to control for confounding. There were
also 3 papers with multivariate regression analysis modelled on less than 100 participants,(50,
92, 102) and 2 papers reporting studies with multivariate regression analysis modelled on more
than 100 patients but with a selected epilepsy population such as excluding patients with
remote symptomatic aetiology(67) and including only patients with focal seizures(103); these
papers were also eliminated.
A total of 119 papers were eliminated, leaving 30 eligible papers. A manual search through the
list of references of each of the eligible papers yielded only 1 potentially eligible title, which
62
turned out to be a case control study without multivariable control for confounding, and so was
eliminated. The full citation details of the papers are in Appendix II.
Figure 6 | Flow diagram of the systematic review
† The number selected after 3 rounds of screening *The 2 papers eliminated from EMBASE were the same as 2 of the 3 eliminated from MEDLINE
Articles identified from electronic MEDLINE search (n=14,967) Articles identified from electronic EMBASE search (n=3,695)
Excluded after reading title and abstract MEDLINE (n=14,818) Excluded after reading title and abstract EMBASE (n=3,675) –Duplication within each database – Review/Editorial/Comments –Non-seizure outcome –Study population restricted to a particular epilepsy subtype –Study population restricted to a particular patient population (e.g. surgical) –Epilepsy or seizure in relation to another pathological process –Epilepsy or seizure as an outcome of another pathological process –First Seizure Studies –AED withdrawal studies –Non English language papers
Potentially eligible from title and abstract MEDLINE (n=149)† Potentially eligible from title and abstract EMBASE (n=20)
Excluded from MEDLINE (n=119), Excluded from EMBASE (n=3) –Cohort Studies without Multivariate Analysis 67 (66 MEDLINE, 1 EMBASE) –Case Control Studies without Multivariate Analysis 4 (MEDLINE) –Studies without Relevant Outcome 44 (MEDLINE) –Multivariate analysis but less than 100 in study 3 (3 MEDLINE, 2 EMBASE)* –Multivariate analysis but selected epilepsy 1 (MEDLINE)
Articles included in Review (n=33) Validation Only Studies – 2 (None from or unique to EMBASE) Cohort Studies with Multivariate Analysis – 25 (2 unique to EMBASE) Nested Case Control Studies with Multivariate Analysis – 6 (1 unique to EMBASE)
Duplication between MEDLINE and EMBASE (n=14)
Eligible from MEDLINE on reading full text (n=30) Eligible from EMBASE on reading full text (n=17)
63
The EMBASE search (from inception to March 2010) returned 3,695 papers, expectedly much
fewer than the broader MEDLINE search. After screening the title and abstracts, 20 papers
remained following the elimination of duplications, review articles, editorials and commentaries,
and papers with non-seizure outcome. Papers reporting studies restricted to a particular
epilepsy subtype or patient population and in which epilepsy or seizure was studied only as an
outcome of or in relation to another pathological process were also eliminated. Reports of first
seizure studies without follow up to the second seizure and beyond and studies of AED
withdrawal were also eliminated as were non-English language papers as in the MEDLINE
search. No additional potentially eligible papers were identified during the second screening of
the search results.
The full text of these 20 papers revealed that there was 1 paper reporting a cohort study but
without multivariate control for confounding. There were also 2 papers with multivariate
regression analysis to control for predictors of seizure outcome, but in fewer than 100
patients.(92, 102) These same 2 papers were also eliminated for the same reason from the
MEDLINE search result. Of the 17 eligible papers from the EMBASE search, 14 had already been
identified from the MEDLINE search, making 47% of the 30 identified from MEDLINE. The
additional EMBASE effort contributed 3 papers of the total 33 (9%), 2 of which were published in
2010. There were no additional papers identified from the list of references of the 3 papers.
(Appendix II)
4.2 DESCRIPTIVE CHARACTERISTICS OF ELIGIBLE PUBLICATIONS Of the 33 publications included, 2 publications exclusively reported on external validation of
prognostic models developed independently in a Dutch cohort and a cohort from Nova Scotia.
The first was a validation of the first Nova Scotia model in patients within the Turku, Finland
cohort.(105) The second was a temporal validation of the first set of models developed from the
Dutch cohort at 2 years follow up.(107)
However, of the other 31 publications, 23 were reports of multivariate regression analysis
conducted on entire cohorts,(21, 24, 32, 63-66, 68-74, 77, 93, 95, 97-101, 104) 7 were case
control analyses of populations nested within cohorts defined by seizure outcome(25, 75, 76,
78, 79, 94, 96) and 1 was an analysis of a combination of 2 existing cohorts,(106) which also
contains an external validation study of prognostic models derived from both cohorts. The study
combined individual patient data from the Dutch and Nova Scotia cohorts, with different
inclusion criteria from the original publications reporting studies on these cohorts to develop a
model from each newly constituted cohort and a third model from the combined population.
The model from each cohort was internally validated on the cohort within which it was
developed, and then cross-validated externally on the other cohort.
64
The general characteristics of the eligible publications are reported in Table 1, showing that
more than 80% were published in the last fifteen years, and almost a third were published in a
single journal, Epilepsia. Only 2 (6%) papers were published in non-neurology journals – New
England Journal of Medicine (NEJM) and Acta Paediatrica (formerly named Acta Paediatrica
Scandinavica). Thirty of the papers were from countries with high income economies, while the
rest are from the 3 contiguous low-and middle- income countries (LMICs) of Bangladesh, India
and Nepal as defined in the World Bank’s 2010 World Development Indicators.(283)
Table 1 | Eligible Publications by Year, Journal and Country
Category
Characteristic
Number
Year of Publication
2006 – 2010 8
2001 – 2005 9
1996 – 2000 10
1991 – 1995 4
1986 – 1990 2
Publishing Journal
Epilepsia 10
Seizure 3
Brain 3
Pediatric Neurology 2
Annals of Neurology 2
Acta Neurologica Scandinavica 2
Others with 1 publication 11
Journal Category
General Medical 1
General Neurological 12
General Paediatric 2
Paediatric Neurology 4
Epilepsy 13
Other (Neurophysiology) 1
Country of Cohort*
Finland 6
Italy 4
Netherlands 4
United Kingdom 2†
USA 6
Canada 3
China (Hong Kong) 2
Others with 1 publication‡ 7
*The sum of this category is 34 instead of 33 as 1 publication is based on 2 cohorts from different countries †The second UK publication is from a multinational cohort with 50% of patients from the UK ‡Four of the 7 countries with 1 publication each are non-European –Bangladesh, India, Nepal and Saudi Arabia; the
others are Norway, Spain and Sweden
65
There were 47 eligible multivariable regression analyses in all, considering different categories
of seizure outcome (Table 2 and Table 3). Each of the 3 publications from low-and middle-
income countries arose from 1 cohort in each of Bangladesh, India and Nepal, with each
publication reporting the only study from each cohort. The same is true for the studies and
cohorts from Hong Kong as well as Saudi Arabia. However, there were 24 eligible studies from
the 10 European cohorts, and 18 studies from 7 North American cohorts. One cohort,(98)
originally from the MESS trial(284) (a randomised controlled trial of the effect of early AED use
on seizure outcome) was multi-national from 9 European countries (United Kingdom, Hungary,
Italy, Netherlands, Poland, Portugal, Slovakia, Turkey and Yugoslavia), 2 South American
countries (Brazil and Chile), India, and Israel. However, as 50% of the cohort was from the
United Kingdom (UK) alone,(284) this study was categorised as European. In all, there was no
cohort from Australasia; neither was there a cohort containing an African population.
Eight (35%) out of a total of 23 unique cohorts were population based cohorts; 5 of the
population based cohorts were European (out of 10) while the remaining 3 were based in North
America (out of 7) (Table 2 and Table 3).
66
Table 2 | Descriptive Characteristics of Cohorts, Publications and Studies (European Cohorts)
Cohort Recruitment Period Setting
Age of Population Decision on AED Regimen
Study Design
Seizure Outcome Method of Analysis (Risk Estimate)
Publication
Turku, Finland 1961-1964, Population
Children (<15 years) Physician’s Discretion Mixed Cohort Remission on or off AED Logistic (OR) Sillanpaa, 1990
Mixed Cohort Remission on or off AED Logistic (OR) Sillanpaa, 1993
Mixed Cohort Remission on or off AED Cox (HR) Sillanpaa, 1998
Mixed Cohort Remission off AED Cox (HR) Sillanpaa, 1998
Prospective Cohort Remission off AED Cox (HR) Sillanpaa, 1998
Prospective Cohort No Remission after Relapse Logistic (OR) Sillanpaa, 2006
Prospective Cohort Remission on or off AED Cox (HR) Sillanpaa, 2009
DSEC, The Netherlands 1988-1992, Hospital
Children (<16 years) Physician’s Discretion Prospective Cohort Remission on or off AED Logistic (OR) Arts, 1999
Prospective Cohort Remission on or off AED Logistic (OR) Arts, 1999
Prospective Cohort Remission on or off AED Logistic (OR) Arts, 2005
Prospective Cohort Remission on or off AED Logistic (OR) Arts, 2005
Prospective Cohort Remission off AED Logistic (OR) Geelhoed, 2005
NGPSE, UK 1984-1987, Population
Children and Adults (>5 years)
Not Stated / Unclear Prospective Cohort Remission on or off AED Cox (HR) MacDonald, 2000
Prospective Cohort Remission on or off AED Cox (HR) MacDonald, 2000
MESS, UK (50%) 1993-2000, Hospital
Children and Adults Randomised Mixed Cohort Early Remission Cox (HR) Kim, 2006
CGSE, Italy 1982-1985, Hospital
Children and Adults Physician’s Discretion Mixed Cohort Early Remission Cox (HR) CGSE, 1988
Mixed Cohort Remission on or off AED Cox (HR) CGSE, 1992
Mixed Cohort Remission on or off AED Cox (HR) CGSE, 1992
Copparo, Italy 1964-1984, Population
Children (<19 years) Not Stated / Unclear Prospective Cohort Intractability Logistic (OR) Casetta, 1999
Bari & Monza, Italy 1989-1999, Hospital
Children and Adults Physician’s Discretion Mixed Cohort Early Remission Logistic (OR) Del Felice, 2010
Västerbotten, Sweden 1985-1987, Population
Adults Physician’s Discretion Mixed Cohort Early Remission Cox (HR) Lindsten, 2001
Akershus, Norway 1987-1994, Population
Adults Not Stated / Unclear Retrospective Cohort Remission on or off AED Logistic (OR) Lossius, 1999
Almeria, Spain 1994-2004, Hospital
Children (<14 years) Physician’s Discretion Mixed Cohort Intractability Logistic (OR) Ramos-Lizana,2009
Mixed Cohort – Combined prospective and retrospective case ascertainment; OR – Odds Ratio; HR – Hazard Ratio
67
Table 3 | Descriptive Characteristics of Cohorts, Publications and Studies (Other Cohorts)
Cohort Recruitment Period, Setting
Age of Population Decision on AED Regimen
Study Design
Seizure Outcome Method of Analysis (Risk Estimate)
Publication
Nova Scotia, Canada 1977-1985, Population
Children (<16 years) Physician’s Discretion Mixed Cohort Remission off AED Logistic (OR) Camfield, 1993
Mixed Cohort Remission off AED Logistic (OR) Camfield, 1993
Mixed Cohort Remission off AED Logistic (OR) Geelhoed, 2005
DSEC and Nova Scotia Children (<16 years) Physician’s Discretion Mixed Cohort Remission off AED Logistic (OR) Geelhoed, 2005
Montreal, Canada 1991-2000, Hospital
Children (2 -17 years) Not Stated / Unclear Retrospective Cohort Remission off AED Cox (HR) Oskoui, 2005
Retrospective Cohort Remission off AED Cox (HR) Oskoui, 2005
Retrospective Cohort Intractability Logistic (OR) Oskoui, 2005
Retrospective Cohort Intractability Logistic (OR) Oskoui, 2005
Retrospective Cohort Intractability Logistic (OR) Oskoui, 2005
Retrospective Cohort Intractability Logistic (OR) Oskoui, 2005
New York, USA 1983-1993, Hospital
Children (<19 years) Not Stated / Unclear Prospective Cohort Early Remission Cox (HR) Shinnar, 2000
Prospective Cohort Remission on or off AED Cox (HR) Shinnar, 2000
Connecticut, USA 1993-1997, Population
Children (<15 years) Protocol Prospective Cohort Intractability Cox (HR) Berg, 2001a
Prospective Cohort Remission on or off AED Cox (HR) Berg, 2001b
Boston, USA Not Stated, Hospital
Children (<18 years) Not Stated / Unclear Retrospective Cohort Intractability Logistic (OR) Ko, 1999
Retrospective Cohort Intractability Logistic (OR) Ko, 1999
Rochester, USA 1935-1978, Population
Children and Adults Not Stated / Unclear Retrospective Cohort Remission on or off AED Cox (HR) Shafer, 1988
New Haven, USA Not Stated, Hospital
Children (Not Stated) Not Stated / Unclear Retrospective Cohort Intractability Logistic (OR) Berg, 1996
Hong Kong, China I 1997-NS, Hospital
Adults Protocol Mixed Cohort Remission on or off AED Logistic (OR) Hui, 2007
Hong Kong, China II 1982-1997, Hospital
Children and Adults Not Stated / Unclear Mixed Cohort Intractability Logistic (OR) Kwong, 2003
Riyadh, Saudi Arabia 1994-1996, Hospital
Adults Protocol Mixed Cohort Remission on or off AED Logistic (OR) Abduljabbar, 1998
Dhaka, Bangladesh 1996-NS, Hospital
Children (<15 years) Protocol Retrospective Cohort Remission on or off AED Logistic (OR) Banu 2003
New Delhi, India Not Stated, Hospital
Children (Not Stated) Not Stated / Unclear Mixed Cohort Intractability Logistic (OR) Chawla, 2002
Kathmandu, Nepal 1995-NS, Hospital
Children and Adults Physician’s Discretion Retrospective Cohort Remission on or off AED Logistic (OR) Lohani, 2010
Mixed Cohort – Combined prospective and retrospective case ascertainment; OR – Odds Ratio; HR – Hazard Ratio
68
4.3 REPORTING CHARACTERISTICS OF ELIGIBLE PUBLICATIONS Table 4 presents the results of 10 reporting characteristics of the 31 eligible publications
reporting multivariate analysis. The publications are assessed based on information documented
within them or within referenced publication(s).
None of the publications reported a power calculation or attempted to justify their sample size
or discuss limitations that may arise as a result of sample size, whereas all the studies reported
on loss to follow up albeit to differing degrees, including studies in which it was reported that no
patient was lost. Only a quarter of the studies mentioned the issue of missing data in the cohort
analysed. However, data completeness was generally poorly described. In several studies,
absence of missing data was a criterion for inclusion in the study analysis. Of the 8 papers that
reported on missing data,(66, 68, 71, 98-100, 104, 106) 3 explicitly stated the number of
participants excluded from analysis for that reason, (68, 104, 106) and 2 reported the use of
imputation to replace missing data.(98, 106)
In their analysis for predictors of seizure outcome, all the studies dichotomised seizure outcome.
However, 3 papers (10%) failed to report the exact proportion or number of participants with
the seizure outcome.(72, 98, 104) The risk estimate or regression coefficient with accompanying
confidence interval or standard error was not reported for all the studies contained in 5
publications.(63, 71, 95, 104, 106) More than half (58%) of the publications reported the
statistical package or software used to conduct the multivariate analysis. The report on the test
of assumptions of multiple regression analysis is as follows: 2 out of 13 (15%) papers using Cox
regression explicitly reported on the test of the assumption proportional hazards.(71, 72)
However, only in 1 paper(65) was there a report of a test for colinearity as a consideration for
including variables in analysis, while 8 papers(24, 65, 66, 73, 96, 98, 101, 104) reported a test of
interaction terms.
Sixteen of the 31 (52%) publications stated the significance level, at which prognostic variables
were retained or excluded from the model, while 2 reported the deliberate inclusion of
prognostic factors considered to be important.(21, 65)Sixteen papers (52%) reported the use of
stepwise selection of variables, backwards, forwards or a combination of both. (Table 4)
69
Table 4 | Reporting characteristics of publications with multivariate analysis
Reporting Characteristic A
bd
ulja
bb
ar, 1
99
8
Art
s, 1
99
9
Art
s, 2
00
4
Ban
u, 2
00
3
Be
rg, 2
00
1a
Be
rg, 2
00
1b
Cam
fie
ld, 1
99
3
CG
SE, 1
98
8
CG
SE, 1
99
2
Ge
elh
oe
d, 2
00
5
Hu
i, 2
00
7
Kim
, 20
06
;
Lin
dst
en
, 20
01
;
Loh
ani,
20
10
Loss
ius,
19
99
Mac
Do
nal
d, 2
00
0
Osk
ou
i, 2
00
5
Ram
os-
Liza
na,
20
09
Shaf
er,
19
88
Shin
nar
, 20
00
Silla
np
aa, 1
99
0
Silla
np
aa, 1
99
3
Silla
np
aa, 1
99
8
Silla
np
aa, 2
00
9
Be
rg, 1
99
6
Cas
etta
, 19
99
Ch
awla
, 20
02
De
l Fe
lice
, 20
10
Ko
, 19
99
Kw
on
g, 2
00
3
Silla
np
aa, 2
00
6
n (
ou
t o
f 3
1)
wit
h ✔
Justification of Study Size
✘
✘
✘ ✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
✘
0/31
Report on Loss to Follow Up
✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ 31/31
Report on N with Outcome
✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✘ ✔ ✔ ✘ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ 29/31
Report of Statistical Package Used
✔ ✔ ✘ ✔ ✘ ✘ ✘ ✔ ✘ ✔ ✔ ✘ ✔ ✔ ✔ ✘ ✔ ✔ ✔ ✘ ✘ ✘ ✔ ✔ ✔ ✘ ✔ ✔ ✔ ✘ ✘ 18/31
Report on/Discussion of Missing Data
✘ ✘ ✔ ✘ ✘ ✘ ✔ ✘ ✘ ✔ ✘ ✔ ✔ ✔ ✔ ✘ ✘ ✘ ✔ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ 8/31
Report on Test for Collinearity
✘ ✔ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ 1/31
Report on Test for Interaction Terms
✘ ✔ ✔ ✘ ✔ ✘ ✘ ✘ ✘ ✘ ✘ ✔ ✘ ✘ ✘ ✔ ✘ ✘ ✔ ✔ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✔ ✘ ✘ ✘ 8/31
Discussion of Model Assumptions
✘ ✘ ✘ ✘ ✘ ✘ ✔ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✔ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ 2/31
Report on Criteria for Variables in Model
✔ ✔ ✘ ✔ ✘ ✔ ✔ ✘ ✘ ✔ ✔ ✔ ✘ ✔ ✔ ✔ ✘ ✔ ✘ ✔ ✘ ✘ ✘ ✔ ✔ ✘ ✘ ✔ ✘ ✘ ✘ 16/31
Report on Results of Analysis
✘ ✔ ✔ ✔ ✔ ✔ ✘ ✘ ✔ ✔* ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✘ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ 27/31
✔ - Yes (Reported); ✘ - No (Not Reported)
* For Geelhoed et al 2005, reporting >1 study where reporting characteristics vary between analyses, the better reporting characteristic is chosen as summary for the paper.
70
4.4 QUALITY APPRAISAL OF ELIGIBLE PUBLICATIONS This segment of the chapter presents the results of quality items on the 31 eligible publications. The
publications are assessed based on information documented within them or within referenced
publication(s).
4.4.1 STUDY PARTICIPATION Every eligible publication had documented within it or within a referenced publication, information on
inclusion criteria and baseline characteristics of the patients included. However, for 7 papers little or no
information on the setting from which the cohort was derived could be retrieved. Four of them provided
insufficient information (78, 93, 96, 100) while the remaining 3 contained no further information beyond
stating the geographical location of the cohort. Two of the 3 were reports from the same cohort (CGSE
in Italy) (64, 95). The third publication was a hospital based nested case control study, (Chawla et al) (94)
and it was the only publication/cohort without the necessary information to make a judgment on
adequate study participation; the judgement was made in the affirmative for all 28 of the other 30
papers.
The studies reported by Ko et al(79)have a clear inversion in the proportion of cases and controls. The
number of patients satisfying the definition of medical intractability in the cohort was more than 3 times
(3.7) as many as those in remission on or off AED. There was also an over-representation of patients
satisfying the criteria for intractability in the study reported by Berg et al 1996.(78) This however may be
explained by the fact that both studies are hospital based retrospective cohorts. It was therefore
concluded that recruitment of eligible individuals into these 2 studies was not adequate.
4.4.2 STUDY ATTRITION Two out of the 24 publications reporting analysis of an entire cohort (as against nested case-control
analysis) conducted their analysis on less than two thirds of the patients in the cohort,(100, 104) owing
mostly to incompleteness of data as they were both retrospective studies. In the case of 1 other
retrospective study,(93) it was uncertain how the cohort was constituted and if there was loss to follow
up.
Two publications with less than 90% completeness of follow up (72, 98) gave details of the
characteristics of patients lost to follow up. In the MESS Trial cohort(98)information from the linked
paper(284)suggest that there were no important clinical differences between those who did not consent
to be randomised into the study and those who consented. However in the second publication,(72)
there was significant difference only in the proportion of patients within the 3 aetiologic groups
(cryptogenic, idiopathic and remote symptomatic). For 4 other publications (reporting studies
conducted on an entire cohort), information upon which to base such judgement was not provided,
even though the studies had >10% participant attrition rate. (69, 77, 100, 104)
71
However, for Chawla et al,(94)a report of a nested case-control study, there was enough reason to
conclude that information on how similar patients included in the analysis compared to those not
included would be important. This was because the study did not provide information about how equal
number (50 in each group) was selected as cases and controls.
4.4.3 PROGNOSTIC FACTORS Twenty three of the 31 publications contained clear definition or description of the prognostic factors
considered for analysis, and only partly so in the remaining 8. However, the handling of continuous
variables was poorly reported, and inferences concerning this were made by a combination of observing
how continuous data were reported and how the authors stated they were handled. In all, only 4 of the
31 publications contained studies in which no continuous variable was categorised.(72, 78, 79, 97)The
poor reporting of missing data reflects on the ability to know if there is adequate proportion of patients
with complete data. This judgement could be made for only 7 publications, 5 (63, 66, 68, 98, 106) of
which had complete data, and 2 (100, 104) that did not have complete data. Also, all the publications
reported on cohorts for which it was unlikely that there would be misclassification bias in assessing the
prognostic factors.
4.4.4 OUTCOME MEASUREMENT As it was for prognostic factor measurement, all the publications reported on cohorts for which it was
unlikely that there is misclassification bias in assessing the chosen outcome measure, as the majority of
patients were followed up for at least 1 year with the outcome clearly defined. The outcome measure
was also clearly defined in all the studies reported in all the publications. Three out of 7 publications
reporting studies with exclusively retrospective ascertainment of cases provided information to
demonstrate that prognostic factors and outcome were assessed blind to each other.(72, 78,
104)However, for Chawla et al,(94) a mixed cohort with retrospective and prospective ascertainment of
cases, the report does suggest that the assessment of the prognostic factors may not have been blind to
outcome assessment at least for some of the patients in the cohort.
4.4.5 MULTIVARIATE ANALYSIS Sixteen papers out of 31 reported the use of some kind of stepwise procedure in the model building
strategy, while only 2 did not report enough data for the results of analysis. Therefore, the calculation of
events per variable ratio was not possible for the 2 publications for which the proportion of patients
with the outcome studied could not be retrieved.(101, 104) It was however less than 10 in studies
reported in 5 publications: 9.8 for 1 of the 2 studies included in Arts et al 2004(66) , 7.5 for the study in
Berg 2001a(73) 8.3 and 4 respectively for 2 categories of study reported by Oskoui et al(72), 9.5 for the
study in Sillanpaa 1993(77)and 7.8 and 6.5 for the 2 studies in Ko et al(79)
72
Table 5 | Details of Quality Appraisal of Publications
Ab
du
ljab
bar
, 19
98
Art
s, 1
99
9
Art
s, 2
00
4
Ban
u, 2
00
3
Be
rg, 2
00
1a
Be
rg, 2
00
1
Cam
fie
ld, 1
99
3
CG
SE, 1
98
8
CG
SE, 1
99
2
Ge
elh
oe
d, 2
00
5
Hu
i, 2
00
7
Kim
, 20
06
;
Lin
dst
en
, 20
01
Loh
ani,
20
10
Loss
ius,
19
99
Mac
Do
nal
d, 2
00
0
Osk
ou
i, 2
00
5
Ram
os-
Liza
na,
20
09
Shaf
er, 1
98
8
Shin
nar
, 20
00
Silla
np
aa, 1
99
0
Silla
np
aa, 1
99
3
Silla
np
aa, 1
99
8
Silla
np
aa, 2
00
9
Be
rg, 1
99
6
Cas
etta
, 19
99
Ch
awla
, 20
02
Del
Fel
ice,
20
10
Ko
, 19
99
Kw
on
g, 2
00
3
Silla
np
aa, 2
00
6
Study Participation
The source population was described Y Y Y P Y Y Y N N Y Y Y Y P Y Y Y Y Y Y Y Y Y Y P Y N P Y Y Y Inclusion criteria and baseline cohort described Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Adequate recruitment of eligible individuals into cohort Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y N Y U Y N Y Y
Study Attrition There is adequate proportion of cohort in study/analysis Y Y Y U Y Y Y Y Y Y Y Y Y N Y Y Y Y N Y P P Y* Y Y Y Y Y Y Y Y Reasons for loss and characteristics described – Y Y U – – – Y Y Y – Y – P – – N N P – Y Y – – – – U – – – – Those lost are similar to those who completed the study – – – U – – – – – – – Y – U – – P – U – U U – – – – U – – – – Prognostic Factors The prognostic variables were clearly defined P P P Y Y Y Y Y Y Y P Y Y Y P Y Y Y Y Y P P Y Y Y P Y Y Y Y Y Continuous variables were kept continuous P P P N P P P P P P Y P N P P P Y P N P P P P P Y P N P Y P P Methods sufficiently limit misclassification bias Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y There was adequate proportion of patients with complete data Y U Y U U U U U U Y U Y U N Y U U U N U U U U U U U U U U U U
Outcome Measurement The outcome measures are clearly defined P Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Measure methods sufficiently limit misclassification bias Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Outcome & prognostic factor measured blind to each other Y Y Y U Y Y Y Y Y Y Y Y Y U U Y Y Y Y Y Y Y Y Y Y Y U Y U Y Y
Multivariate Analysis There is sufficient data to assess adequacy of the analysis Y Y Y Y Y Y Y N Y Y Y Y Y Y Y Y Y Y N Y Y Y Y Y Y Y Y Y Y Y Y The model building strategy is described as being stepwise Y Y N Y N N Y Y N Y N Y N N Y Y N N Y N Y Y N N N N Y Y N Y Y There is no over-fitting the data (i.e. EPV at least 9) Y Y Y* Y N Y Y Y Y Y* Y Y Y Y Y Y* Y* Y Y Y Y Y Y Y Y Y Y Y N Y Y
Y-Yes, P-Partly, U-Unsure, N-No | Dash (–) is used when the item is not stated, while it is either not applicable or unimportant as a potential source of bias in the cohort/study being reported.
*For papers reporting >1 study where result of quality appraisal vary between studies, (Geelhoed et al, Oskoui et al and Sillanpaa et al 1998) the quality of the better study is reported in this table.
EPV – Events per Variable
73
4.5 STUDY CATEGORIES BY OUTCOME MEASURE The next segment of this chapter considers each study within designated categories, classified
according to the seizure outcome predicted in the study in order to facilitate comparability
among studies as different seizure outcome measures were considered across studies. The 5
categories (Type I-V) of studies are introduced. The results are subsequently reported in relation
to quality considering the following – 1) Proportion of patients with outcome; 2) Predictors of
the outcome; 3) Non predictors of the outcome; and 4) Prognostic models for predicting the
outcome
STUDIES PREDICTING EARLY/IMMEDIATE REMISSION (TYPE I) Type I Studies are those that identified independent predictors of belonging or not to Sub-Group
1A, which is the category of patients with immediate or early spontaneous remission (ISR).
These patients will not have a third seizure, i.e. upon starting AED or being recruited into the
cohort under consideration they enter terminal remission immediately, and will not relapse
after complete AED withdrawal. These studies (24, 95, 98, 99) typically include the entire cohort
i.e. all the patients with early or immediate spontaneous remission (ISR) compared with those
who did not enter ISR. (Figure 4)There was however 1 atypical study, a nested case-control
study, by Del Felice et al(96) that was classified within this category. The study compares
patients with early or immediate spontaneous remission (ISR) i.e. no additional seizure once
AED therapy started with “late” remission i.e. those who continued to have seizures until 2 years
after AED therapy was commenced with those patients. (Figure 7)
The colour differences (lemon and grey) in the lowest set of boxes indicate comparison groups.
STUDIES PREDICTING REMISSION OFF AED (TYPE II) Type II Studies, illustrated in Figure 8, identified independent predictors of remission off AED,
i.e. patients in prognosis Group 1. This group consists of patients in Sub-Group 1A (immediate
spontaneous remission – ISR) and Sub-Group 1B with non-immediate spontaneous remission
(NISR) i.e. patients who although do not enter remission immediately, ultimately do so, and
remain in remission upon AED withdrawal. The other comparison group therefore is all the
other patients on AED whether in remission or not.(32, 71, 72, 106)
PEOPLE WITH EPILEPSY
REMISSION
1. REMISSION OFF AED
1A. NO 3RD SEIZURE 1B >2 SEIZURES
2. REMISSION ONLY ON AED
2A NO PRIOR RELAPSE 2B PRIOR RELAPSE(S)
NO REMISSION
3.NO REMISSION ON OR OFF AED
3A RELAPSE OR NO RELAPSE
3B INTRACTABLE
Figure 7 | Type I studies (Predicting Immediate Remission)
74
Figure 8 | Type II studies (Predicting Remission off AED)
The colour differences (lemon and grey) in the lowest set of boxes indicate comparison groups.
STUDIES PREDICTING REMISSION ON OR OFF AED (TYPE III) Type III Studies consider as an outcome group, those patients in remission, on or off AED. These
patients belong in the prognosis Group 1 and Group 2. Therefore the other comparison group
for this category of studies are those patients who do not enter remission either for a specified
period during the course of follow up or immediately before the last follow up assessment. This
is illustrated in Figure 9, and shows how the second comparison group contains patients with
intractable seizures, and those who although are not in remission, do not meet the criteria for
intractability.(24, 32, 63-66, 68-70, 77, 93, 97, 100, 101, 104)
Figure 9 | Type III studies (Predicting Remission on or off AED)
The colour differences (lemon and grey) in the lowest set of boxes indicate comparison groups.
STUDIES PREDICTING INTRACTABILITY (TYPE IV) These are studies predicting patients that will have medically intractable seizures. There are 2
kinds of Type IV Studies identified as eligible. There are those where patients satisfying the
definition of medical intractability (Sub-Group 3B) are compared with all other patients who do
not (Group 1, Group 2 and Sub-Group 3A). These studies include all the patients within the
entire cohort (72-74) as shown in Figure 10.
PEOPLE WITH EPILEPSY
REMISSION
1. REMISSION OFF AED
1A. NO 3RD SEIZURE 1B >2 SEIZURES
2. REMISSION ONLY ON AED
2A NO PRIOR RELAPSE 2B PRIOR RELAPSE(S)
NO REMISSION
3.NO REMISSION ON OR OFF AED
3A RELAPSE OR NO RELAPSE
3B INTRACTABLE
PEOPLE WITH EPILEPSY
REMISSION
1. REMISSION OFF AED
1A. NO 3RD SEIZURE 1B >2 SEIZURES
2. REMISSION ONLY ON AED
2A NO PRIOR RELAPSE 2B PRIOR RELAPSE(S)
NO REMISSION
3.NO REMISSION ON OR OFF AED
3A RELAPSE OR NO RELAPSE
3B INTRACTABLE
75
Figure 10 | Type IV studies (Predicting Intractability including the entire cohort)
The colour differences (lemon and grey) in the lowest set of boxes indicate comparison groups.
The other kind of study are case control studies nested within cohorts that compare only
patients with remission on or off AED (Group 1 and Group 2) being controls with patients
satisfying the definition of intractability (Sub-Group 3B) as cases, leaving out those patients in
Sub-Group 3a who are not in remission but also do not fulfil the criteria for medical
intractability.(75, 76, 78, 79, 94) This is illustrated in Figure 11, with Sub-Group 3A taken off.
Figure 11 | Type IV studies (Predicting Intractability in nested case control studies)
The colour differences (lemon and grey) in the lowest set of boxes indicate comparison groups.
STUDIES PREDICTING NO REMISSION AFTER RELAPSE (TYPE V) The last type of study, Type V Studies, has only 1 study within the category. This study reported
by Sillanpaa and Schmidt 2006(25) considers the predictors of not achieving remission on or off
AED after a patient’s seizures have run a remitting-relapsing course. Sub-Group 3A represents
patients without remission that do not satisfy the criteria for intractability. However, although
there are also within the group those who have never experienced remission, Figure 12 assumes
that Sub-Group 3A sufficiently represents those who have not achieved remission after a
remitting-relapsing course.
PEOPLE WITH EPILEPSY
REMISSION
1. REMISSION OFF AED
1A. NO 3RD SEIZURE 1B >2 SEIZURES
2. REMISSION ONLY ON AED
2A NO PRIOR RELAPSE 2B PRIOR RELAPSE(S)
NO REMISSION
3.NO REMISSION ON OR OFF AED
3A RELAPSE OR NO RELAPSE
3B INTRACTABLE
PEOPLE WITH EPILEPSY
REMISSION
1. REMISSION OFF AED
1A. NO 3RD SEIZURE 1B >2 SEIZURES
2. REMISSION ONLY ON AED
2A NO PRIOR RELAPSE 2B PRIOR RELAPSE(S)
NO REMISSION
3.NO REMISSION ON OR OFF AED
3A RELAPSE OR NO RELAPSE
3B INTRACTABLE
76
Figure 12 | Type V studies (Predicting no remission after relapse)
The colour differences (lemon and grey) in the lowest set of boxes indicate comparison groups.
4.6 REVIEW OF STUDIES BY OUTCOME CATEGORY The studies are now reviewed within these categories, especially in relation to: 1)
Understanding and explaining similarities and differences in results given each study’s peculiar
characteristics and potential for bias, 2) Identifying consistent independent predictors and non-
predictors within each category, and 3.) Assessing the quality and performance of externally
validated prognostic models within each outcome category.
PREDICTING EARLY/IMMEDIATE REMISSION (TYPE I) Five studies investigated the predictors of early or immediate remission (Table 6).(24, 95, 96, 98,
99) One of the studies, Del Felice et al,(96) was a nested case control study with patients
achieving remission after at least 2 years of antiepileptic drug therapy as cases(11% of the entire
cohort), and having patients with immediate remission as controls, i.e. those without seizure
recurrence once started on antiepileptic drugs, and who remained in remission for 2 years (33%
of the entire cohort).
In the remaining 4 studies, the proportion of patients without seizure recurrence at 2 years of
follow up (i.e. early or immediate remission) ranged from 37% in Shinnar et al,(24) (the only
study with exclusively prospective case ascertainment) to between 42% and 57.5% in studies
with mixed case ascertainment, including different proportions of first seizure, prospectively
identified, and retrospectively ascertained cases.
Two of the studies, CGSE 1988 (95) and Del Felice 2010 (96), both in all age cohorts, had
remarkably similar proportion of patients in each seizure category with the ratio of 1st seizure
to 2-5 seizures and >5 seizures being roughly 4:9:7 in both studies. In spite of this similarity, the
proportion of patients that achieved immediate remission at about 2 years of follow up varied
widely: 48% in CGSE 1988,(95) and 33% in Del Felice 2010.
Table 6 presents potential predictors that were examined by at least 2 studies. Neither age nor
gender was found to be significantly related to achieving early/immediate remission.
PEOPLE WITH EPILEPSY
REMISSION
1. REMISSION OFF AED
1A. NO 3RD SEIZURE 1B >2 SEIZURES
2. REMISSION ONLY ON AED
2A NO PRIOR RELAPSE 2B PRIOR RELAPSE(S)
NO REMISSION
3.NO REMISSION ON OR OFF AED
3A RELAPSE OR NO RELAPSE
3B INTRACTABLE
77
The number of seizures (as a continuous variable), was only significant in 2 studies in the
univariate analysis. However, the 2 studies found having “2 or more” seizures before the index
seizure or AED was retained in their multivariable models (Box 1):
All 5 studies considered the early or intake Electro Encephalogram (EEG). In 3, EEG finding was
not associated with early remission on univariate analysis. One found that EEG finding was
associated, and in 1 study, having an abnormal EEG was retained in the multivariable model.
Neither a family history of epilepsy nor a prior history of febrile seizures was found to be
predictive of early or immediate remission.
The 5 studies included also considered having aetiological factors as potential predictors, with
three of the 5 retaining the variable in their multivariable models:
None of the models were externally validated.
Kim et al 2006 | RR 0.74 (95%CI 0.58-0.94) Shinnar et al 2000 | RR 0.59 (95%CI 0.41-0.86) Lindsten et al 2001 | RR 0.44 (95% CI 0.26-0.77)
Lindsten et al 2001 | RR 0.63 (95%CI 0.36-1.11) CGSEPI 1988 | (Not Stated)
Box 1 | Relative risk (RR) of achieving early remission in patients with two or more seizures before intake
Box 2 | Relative risk (RR) of achieving early remission in patients with remote symptomatic aetiology
78
Table 6 | Studies predicting immediate remission with information on potential for bias
Age* (Years)
Study (Ref)
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure % with Outcome•
Independent Predictors
Risk Estimate (95%CI)
0 – 19 Shinnar, 2000(24)
Prospective Hospital N=182
Y Y Y Y Y Excellent case ascertainment – NDE‡ (100%)
8.4yrs (Mean)
No 3rd Seizure (37% at 2 years, 28% at 5 years)
Remote Symptomatic Aetiology AED within 3mo of 2nd
Seizure
AED after 3mo of 2nd Seizure 2nd Seizure within 3months 2nd Seizure between 3-6months
HR: 1.69 (1.16-2.47) HR: 0.37 (0.22-0.63) HR: 1.28 (0.75-2.18) HR: 4.00 (2.28-7.00) HR: 0.86 (0.49-1.50)
2 – 81 (19) CGSE, 1988(95)
Mixed† Hospital N=283
Y Y Y Y Y Progressive neurological disorders excluded 1st Seizure (18%), 2-5 Seizures (46%), >5 Seizures 34%)
0.1-3.3yrs Next seizure after Intake (48% at ≈2.5 years)
≥2 seizures before intake Mixed seizure types
NS NS
IQR (17.4-43.4)
Kim, 2006(98)
Mixed† Hospital N=1443
Y Y Y Y Y RCT, therefore known poor prognosis excluded. 1st seizure (57%), 2nd seizure – NDE‡ (24%), and 3 or more seizures – retrospective case ascertainment (19%)
3.0-6.3yrs Next seizure after Intake (57.5% at 2 years, 47% at 5 years, 43.5% at 8 years)
Log No seizures before intake Abnormal EEG Neurological Disorder
HR:1.56 (1.42-1.72) HR: 1.54 (1.27-1.86) HR: 1.35 (1.07-1.72)
17-83 Lindsten, 2001(99)
Mixed† Population N=107
Y Y N Y Y 1st seizure (29%), 2nd seizure – NDE‡ (15%), and 3 or more seizures – retrospective case ascertainment (56%)
8.8-13.5yrs Next seizure after Index (42%at 2 years)
≥2 seizures before intake No AED Therapy Remote Symptomatic Aetiology
HR: 1.6 (0.9-2.8) HR: 0.29 (0.11-0.74) HR: 2.25 (1.3-3.9)
3-84 (31.5) Del Felice, 2010(96)
Mixed† Hospital N=352 Cases 38 Controls 115
Y Y Y Y Y 1st Seizure (16%), 2-5 Seizures (46%), >5 Seizures 38%) This is a case control study nested within a cohort
6.9yrs (Mean)
Cases 2 year seizure- freedom begins at least 2 years after AED therapy starts Controls 2 year seizure- freedom begins with AED (33% at 2 years)
2–5 Partial Seizures before AED >5 Partial Seizures before AED
OR: 2.7 (1.0-6.8) OR: 6.7 (2.3-19.3)
*Age range in years (Mean ± Standard Deviation); •The proportion reported is of those who did not have a relapse following diagnosis (2nd seizure), recruitment or commencement of AED; †Mixed–Cohort combined prospective and retrospective identification of cases; ‡NDE – Newly Diagnosed Epilepsy i.e. cases that were ascertained on having the second seizure. Case ascertainment after 2nd seizure is considered retrospective; Y–Yes, N–No, CI-Confidence Interval, CI-Confidence Interval, IQR–Interquartile Range, RCT–Randomised Controlled Trial, AED–Antiepileptic Drug, EEG–Electro Encephalogram, No–Number, NS–Not Stated, OR–Odds Ratio, HR–Hazard Ratio
79
Table 7 | Consistent predictors and non-predictors of immediate remission
Shin
nar
, 20
00
CG
SE, 1
98
8
Kim
, 20
06
Lin
dst
en
, 20
01
De
l Fe
lice
, 20
10
Demographic
Age at Onset ✘ ✘
Gender ✘ ✘ ✘
Epilepsy Before AED, Intake or Index
Duration ✘ Θ ✘ ✘
≥2 seizures before Intake ✔ ✔
N of Seizures before Intake ✔ ✔
EEG
Intake/Early - EEG Abnormal ✘ ✘ ✔ ✘ ✔
Seizure Type
Generalised Onset ✘ ✔
Partial ✘ ✔
Aetiology | Syndrome
Remote Symptomatic Aetiology ✔ ✘ ✔ ✔ ✘
Neurological Signs
Neurological Examination ✘ ✘
Others
Family History ✘ Θ ✘
Febrile Seizure ✘ ✘
✔ - Variable is significant or retained in multivariate model
✔ - Variable is only significant on univariate analysis
✘- Variable is not significant on univariate analysis Θ – Variable is not reported in univariate analysis, but reported as not significant on multivariate analysis
80
PREDICTING REMISSION OFF ANTIEPILEPTIC DRUGS (TYPE II) Nine studies considered the predictors of remission off antiepileptic drugs. They were all conducted
in childhood cohorts. Three were from the Nova Scotia cohort,(71, 106) 2 from each of the
Turku,(32) and Montreal cohorts,(72) 1 from the Dutch cohort,(106) and the last one was an
analysis of reconstituted and combined data from the Nova Scotia and Dutch cohorts.(106) (Tables
8 and 9)
The proportion of patients that remained in remission while successfully weaned off antiepileptic
medication ranged from 52% in a retrospective cohort to between 47% and 55% in mixed cohorts
and from 56% to 65% in prospective cohorts. It was 59% in the combined cohort of the
reconstituted (after including patients with known poor prognosis and generalised absence
seizures) Nova Scotia cohort (55%) and the Dutch DSEC cohort (65%).
The potential predictors considered for remission off antiepileptic drugs and found to be consistent
predictors or non-predictors are presented in Table 10. Age at onset of seizures and gender were
not found to be significantly related to remaining in remission while taken off antiepileptic
medication. However, having more than 1 seizure in the period between 6 and 12 months while on
AED was assessed in the 2 studies that considered onset variables in combination with variables
assessed after 1 year of AED medication,(71, 72)and was found to consistently predict remission off
AED. The relative risk was 0.24 (95% CI 0.10-0.60) from Oskoui et al,(72) while the relative risk is
similar at 0.25 (95% confidence intervals not provided) from Camfield et al(71).
Cognitive impairment at onset or diagnosis of epilepsy was another consistent predictor of
remission off AED, considered in all the 9 models. It was retained in 6 of the 9, and excluded in the
remaining three studies, which however did not report the test of univariate association. The
relative risk from 2 of the three studies using only intake variables, diverged widely, albeit with
overlapping confidence intervals, at 0.77 (95% CI 0.61-0.94) from the combined Nova Scotia and
Dutch cohort [Geelhoed et al(106)] and 0.23 (95%CI 0.07-0.74) from the Montreal cohort [Oskoui
et al(72)]. The third study, Camfield et al,(71) did not report 95% confidence interval. The Geelhoed
et al cohort(106) includes patients from the Camfield et al cohort.(71)
Oskoui et al 2005 |RR 0.24 (95%CI 0.10-0.60) Camfield et al 1993 |RR 0.25 (95% CI Not Stated)
Oskoui et al 2005 | RR 0.23 (95%CI 0.07-0.74) Geelhoed et al 2005 | RR 0.77 (95%CI 0.61-0.94) Camfield et al 1993 |RR 0.25 (95%CI Not Stated)
Box 3 | Relative risk (RR) of remission off AED in children having more than 1 seizure 6 to 12 months on AED
Box 4 | Relative risk (RR) of remission off AED in children with intellectual disability (models with only onset variables)
81
Table 8 | Studies predicting remission off AED (Nova Scotia and DSEC Cohort)
Age (Years)*
Study
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure (% with Outcome)
Independent Predictors
Risk Estimate (95%CI)
0-16 (6.7±4.5)
Nova Scotia Camfield, 1993(71)
Mixed Population N=486
Y Y Y Y Y Patients with known poor prognosis, generalised absence and progressive neurological disorders excluded Only intake variables
7.1yrs (Mean)
Off AEDs at Last Follow up (54%)
Age ≤ 1yr Age 1-12yrs Intellectual disability History of Neonatal Seizures 1-2 Seizures before AED ≥3 Seizures before AED
HR: 0.23 (NS) HR: 0.21 (NS) HR: 4.00 (NS) HR: 0.17 (NS) HR: 0.42 (NS) HR: 0.31 (NS)
0-16 (6.7±4.5)
Nova Scotia Camfield, 1993(71)
Mixed Population N=486
Y Y Y Y Y Patients with known poor prognosis, generalised absence and progressive neurological disorders excluded Intake and 1yr variables
7.1yrs (Mean)
Off AEDs at Last Follow up (54%)
Age ≤ 1yr Age 1-12yrs Intellectual disability History of Neonatal Seizures 1-2 Seizures before AED ≥3 Seizures before AED ≤1 Seizure 6-12months on AED >1 Seizure 6-12months on AED
HR: 0.37 (NS) HR: 0.17 (NS) HR: 3.45 (NS) HR: 0.16 (NS) HR: 0.42 (NS) HR: 0.31 (NS) HR: 2.78 (NS) HR: 4.00 (NS)
0-16 Nova Scotia Geelhoed, 2005(106)
Mixed Population N=602
Y Y Y Y Y Includes all epilepsy types from the Nova Scotia cohort
8.8yrs (Mean)
Off AEDs at Last Follow up (55%)
Idiopathic Partial Epilepsy Cryptogenic Partial Epilepsy Intellectual disability
NS NS NS
0-16 (5.9±4.2)
DSEC Cohort Geelhoed, 2005(106)
Prospective Hospital N=453
Y Y Y Y Y Exact same cohort as reported in Arts et al, 2004
5yrs (Minimum)
Off AEDs at Last Follow up (65%)
Age >12yrs Symptomatic Gen. Epilepsy Cryptogenic Gen. Epilepsy Symptomatic Partial Epilepsy Cryptogenic Partial Epilepsy
NS NS NS NS NS
0-16 Nova Scotia & DSEC Cohort Geelhoed, 2005(106)
Combination Mixed (Hospital and Population) N=1055
Y Y Y Y Y Includes all epilepsy types from Nova Scotia cohort merged with the Dutch DSEC cohort
5yrs (Minimum for >95%)
Off AEDs at Last Follow up (59%)
Age >12 yr N seizures before AED Neurologic Deficits Intellectual disability Absence Seizures Cryptogenic Gen. Epilepsy Symptomatic Partial Epilepsy Cryptogenic Partial Epilepsy + History of Febrile Seizures
OR: 2.04 (1.67, 2.63) OR: 1.02 (1.01, 1.04) OR: 1.61 (1.04, 2.5) OR: 1.81 (1.19, 2.28) OR: 0.57 (0.36, 0.95) OR: 2.17 (1.22, 3.85) OR: 2.94 (1.92, 4.34) OR: 1.75 (1.33, 2.38)
*Age range in years (Mean ± Standard Deviation); CI-Confidence Interval, Y-Yes, HR-Hazard Ratio, OR-Odds Ratio, AED-Antiepileptic Drug, NS-Not Stated
82
Table 9 | Studies predicting remission off AED (Montreal and Turku Cohort)
Age* (Years)
Study
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure (% with Outcome)
Independent Predictors
Risk Estimate (95%CI)
0-15 (4.3)
Turku, Fin Sillanpaa, 1998(32)
Mixed Population N=176
Y U Y Y Y Three seizures diagnosed epilepsy
23-39yrs Off AEDs 5yrs before Last Follow up (47%)
75-100% reduction in seizures within 3months of AED Complex Partial Seizures Cryptogenic vs. R. Symptomatic Idiopathic vs. R. Symptomatic
HR: 0.09 (0.03, 0.31) HR: 3.03 (1.40, 6.67) HR: 0.33 (0.15, 0.71) HR: 0.32 (0.16, 0.65)
0-15 Turku, Fin Sillanpaa, 1998(32)
Prospective Population N=117
Y Y Y Y Y Three seizures diagnosed epilepsy Only prospectively identified patients in Turku cohort
11-39yrs Off AEDs 5yrs before Last Follow up (56%)
75-100% reduction in seizures within 3months of AED Complex Partial Seizures Atonic Seizures Cryptogenic vs. R. Symptomatic Idiopathic vs. R. Symptomatic
HR: 0.45 (0.25, 0.81) HR: 3.57 (1.70, 7.69 ) HR: 3.85 (0.83, 16.68) HR: 0.30 (0.14, 0.64) HR: 0.38 (0.18, 0.83)
2-17 (7.6±3.7)
Montréal, Ca Oskoui, 2005 (72)
Retrospective Hospital N=196
Y P Y Y Y Only onset variables Data collection was remarkably good
2-13.6yrs Off AEDs at Last Follow up (52%)
Intellectual disability Remote Symptomatic Epilepsy Mixed Seizure Types
HR: 4.35 (1.35, 14.29) HR: 2.17 (1.09, 4.35) HR: 2.94 (1.28, 6.67)
2-17 (7.6±3.7)
Montréal, Ca Oskoui, 2005(72)
Retrospective Hospital N=196
Y P Y Y Y Onset and 1yr variables Data collection was remarkably good
2-13.6yrs Off AEDs at Last Follow up (52%)
Intellectual disability >1 Seizure 6-12months on AED Mixed Seizure Types
HR: 5.26 (1.70, 16.68) HR: 4.17 (1.67, 10.0) HR: 2.5 (1.08, 5.56)
*Age range in years (Mean ± Standard Deviation); CI-Confidence Interval, Y-Yes, Unclear, P-Partly, HR-Hazard Ratio, OR-Odds Ratio, AED-Antiepileptic Drug
83
Table 10 | Consistent predictors and non-predictors of remission off AED
Cam
fie
ld, 1
99
3
Cam
fie
ld, 1
99
3
Ge
elh
oe
d, 2
00
5N
Ge
elh
oe
d, 2
00
5D
Ge
elh
oe
d, 2
00
5N
D
Silla
np
aa, 1
99
8
Silla
np
aa, 1
99
8†
Osk
ou
i, 2
00
5
Osk
ou
i, 2
00
5*
Demographics
Age at Onset Θ Θ Θ Θ
Gender Θ Θ Θ Θ Θ
Epilepsy Before AED, Intake or Index
N of Seizures before Intake or AED ✔ Θ Θ Θ Θ
Early Epilepsy Characteristics
>1 Seizure from 6 - 12months ✔ ✔
EEG
Intake/Early - EEG Abnormal ✔ Θ Θ Θ Θ Θ
Aetiology | Syndrome
Remote Symptomatic Aetiology ✔ Θ ✔ Θ
Cryptogenic Epilepsy Θ ✔ ✔ Θ Θ
ILAE Syndrome Θ Θ Θ Θ
Cognition
Abnormal Cognitive Development ✔ ✔ ✔ Θ ✔ Θ Θ ✔ ✔
Neurological Sign
Neurological Examination ✔ Θ Θ ✔ Θ Θ Θ Θ
✔ - Variable is significant or retained in multivariate model
✔ - Variable is only significant on univariate analysis
✘- Variable is not significant on univariate analysis
Θ – Variable is not reported in univariate analysis, but reported as not significant on multivariate analysis Intellectual disability was also retained in 2 studies (71, 72) that combined variables assessed at
intake or onset with those only assessable at 1 year after onset/intake (e.g. number of seizures
between 6-12 months while on AED):
None of the studies that considered abnormal EEG at intake or early in the course of epilepsy
confirmed it as an independent predictor of remission off AED. However, 1 study (106) out of 5
found the number of seizures before intake to be an independent predictor. The same model
(106)also retained deficit on neurological examination as an independent predictor out of 8
studies that considered signs on neurological examination.
Oskoui et al 2005 | RR 0.19 (95%CI 0.06, 0.59)
Camfield et al 1993 |RR 0.29 (95%CI Not Stated)
Box 5 | Relative risk (RR) of remission off AED in children with intellectual disability (models combining onset and 1 year variables)
84
Three of the models were externally validated. The results of the validation studies are
presented in Table 11. Sillanpaa et al(105) externally validated the Nova Scotia model
(developed using the intake variables only) on the Turku cohort, although this was to predict
three years seizure-free status immediately before last follow up on or off medication instead of
remission off AED. This was because physicians handling patients within the Turku cohort were
reluctant to discontinue medication and as 75% of those with 3 year remission at last follow up
had been in remission for at least 10 years, it was considered unlikely that these patients would
relapse off medication.
Patients in the Turku cohort were selected with the same specific inclusion criteria in the Nova
Scotia cohort for the validation study. The performance of the Nova Scotia intake model(105)
had greater specificity (88% vs. 64%) and positive predictive value (84% vs. 71%) compared to
the results of internal validation, but much poorer sensitivity (43% vs. 73%) and negative
predictive value (51% vs. 67%), and fewer instances of correct prediction (61% vs. 67%)
The other 2 externally validated models were reported by Geelhoed et al(106), from the model
derived from the reconstituted Nova Scotia cohort and from the DSEC cohort to predict
remission off AED, externally validated in the other, with similar results on external and internal
validation in both models.
85
Table 11 | Externally validated models predicting remission off AED
Predictors in Model
Derivation Cohort
Internal Validation: Performance of Model in Derivation Cohort
Validation Study
Validation Cohort
External Validation: Performance of Model in Validation Cohort Sc
ore
Easy
to
Use
Imp
ort
ant
Var
iab
le
Pro
bab
ility
or
Act
ion
Clin
ical
Use
Stu
die
d
Age at Onset -< 1 yr, -1-12 yrs, ->12 yrs Intelligence -Normal/-Retardation Neonatal Seizure No/Yes Seizures before AED -1 or 2, -3 to 20, ->20
Camfield 1993(71) Nova Scotia Patients with known poor prognosis, generalised absence and progressive neurological disorders excluded Pre-Test Probability (54%)
AUC (NS) Best probability cut-off (NS) Sensitivity (74% ± 4%) Specificity (64% ± 5%) PPV (71% ± 4%) NPV (67% ± 5%) Correct Prediction (68%)
Sillanpaa 1995(105)
Turku Cohort 141 patients with the same inclusion criteria as the Nova Scotia Cohort. The Turku cohort had significantly more patients with intellectual disability Validated predicting remission, and not remission off AED Pre-Test Probability (60%)
AUC (NS) Probability cut-off (NS) Sensitivity (43%) Specificity (88%) PPV (84%) NPV (51%) Correct Prediction (61%)
Y Y Y Pr N
Idiopathic Partial Epilepsy Cryptogenic Partial Epilepsy Intellectual disability
Geelhoed 2005(106) Reconstituted Nova Scotia Cohort to includes all epilepsy types Pre-Test Probability (55%)
AUC (NS) Best probability cut-off (≈50%) Sensitivity (69%) Specificity (69%) PPV (73%) NPV (35%) Correct Prediction (69%)
Geelhoed 2005(106)
Geelhoed 2005 DSEC Dutch Cohort Validation cohort similar to derivation cohort Pre-Test Probability (65%)
AUC (NS) Probability cut-off (50%) Sensitivity (71%) Specificity (58%) PPV (76%) NPV (53%) Correct Prediction (64%)
N N Y Pr N
Age >12yrs Symptomatic Gen. Epilepsy Cryptogenic Gen. Epilepsy Symptomatic Partial Epilepsy Cryptogenic Partial Epilepsy
Geelhoed 2005(106) DSEC Dutch Cohort Pre-Test Probability (65%)
AUC (NS) Best probability cut-off (≈50%) Sensitivity (70%) Specificity (66%) PPV (79%) NPV (55%) Correct Prediction (69%)
Geelhoed 2005(106)
Geelhoed 2005 Reconstituted Nova Scotia Cohort Validation cohort similar to derivation cohort Pre-Test Probability (55%)
AUC (NS) Probability cut-off (50%) Sensitivity (57%) Specificity (67%) PPV (67%) NPV (56%) Correct Prediction (61%)
N N N Pr N
AUC-Area Under Receiver Operator Characteristics (ROC) Curve, PPV-Positive Predictive Value, NPV-Negative Predictive Value, NS-Not Stated, Pr-Probability, N-No, Y-Yes, AED-Antiepileptic Drug
86
PREDICTING REMISSION ON OR OFF AED (TYPE III) Twenty studies (Table 12 to 14) investigated the independent predictors of achieving remission
irrespective of whether patients continue on antiepileptic medication or are taken off AEDs
successfully. Eleven of the studies were analysed using logistic regression analysis with the odd
ratio for developing refractory seizures (i.e. not entering remission on or off AED) as risk
estimate, whereas studies using the Cox regression analysis, estimated the relative risk for
achieving remission. However, the Nepalese study reported by Lohani et al(100)was not
conducted among patients with newly diagnosed epilepsy as 50% had seizures for more than 1
year before intake into the cohort and antiepileptic medication.
The proportion of patients in remission ranged from as low as 50% in a retrospective cohort (93)
to as high as 76% in 2 prospective studies (66, 70) and even higher at 80% in a mixed cohort (63)
with the proportion of patients in remission varying widely within the same study design
according to how remission was defined: studies that defined remission as “seizure-free period
immediately before last assessment” (terminal remission) tend to have less proportion in
remission than studies that defined remission simply as “seizure-free period during follow up.”
For example, in 2 adult-onset epilepsy mixed cohorts with 1 year minimum follow up, the study
that defined remission as 1 year terminal had 60% in remission,(97) while the study that defined
remission as 1 year “seizure free period during follow up” had 80%. (63) However, in prospective
studies more patients were in remission than in mixed or retrospective studies with the same
outcome definition and minimum follow up period. (Tables 15 and 16)
Having mixed (multiple) seizure types at onset was found to predict remission in studies from 2
independent cohorts. The Italian CGSE cohort retained mixed seizure types in the model
predicting remission as 2 years seizure-free status during follow up [relative risk 0.70 (95% CI
0.37-1.04)]and 3 years seizure-free status [relative risk 0.70 (95%CI 0.37-1.04)] by 5 years of
follow up.(64) Banu et al (93) found that the odds of achieving remission (3 months terminal
remission at I year minimum follow up) was about 0.23 (95% CI 0.10-0.48) times in patients with
multiple seizure types at onset compared to those with single seizure types. However, Banu et
al(93) did not provide sufficient data to assess the absolute risk of not achieving remission in
patients without multiple seizure types in order to compute the relative risk from odds ratio.
*RR-Relative Risk, •OR-Odds Ratio
CGSEPI 1992 | RR* 0.70 (95%CI 0.37-1.04) Banu et al 2002 | OR• 0.23 (95% CI 0.10-0.48)
Box 6 | Risk of achieving remission in patients with mixed seizure types at onset
87
Four studies from different cohorts found remote symptomatic aetiology to be an independent
predictor of remission with similar risk estimates.(21, 24, 77, 104)
Banu et al(93) and Hui et al(97) also identified intellectual disability as an independent predictor
of remission. However the outcome measure used in the cohort reported by Banu et al(93) [3
months terminal remission] is not comparable to that used by Hui et al(97) [1 year terminal
remission]This difference in outcome measure, together with the disparity in the proportion of
patients in remission – 50% (Banu et al) and 60% (Hui et al) – and possible difference in the
definition of intellectual disability (Hui et al did not specify how this was defined), and the fact
that Banu et al(93) was a childhood cohort while Hui et al (97) was in adults may be responsible,
at least in part, for the wide disparity in the risk estimates:
The frequency of seizures, before and after intake or AED, albeit defined differently in studies,
was also found to be a consistent predictor of remission. The model fitted by Arts et al 1999(65)
with only intake variables, retained the natural logarithm transformation of the number of
seizures before intake as an independent predictor. The model reported by MacDonald et
al(101)to predict 1 year seizure-free period during follow up also retained the log
transformation of the number of seizures before intake. The estimate of the relative risk in Arts
et al 1999 (predicting 6 months terminal remission) is compared with that from MacDonald et al
(predicting experiencing 1 year free of seizures during follow up):
Shinnar et al 2000 | RR 0.47 (95% CI 0.27- 0.81) Berg et al 2001b | RR 0.63 (95% CI 0.47-0.84) Sillanpaa 1993 | RR 0.44 (95% CI 0.25-0.92) Shafer et al 1988 | RR 0.44 (95% CI Not Stated)
Box 7 | Relative risk (RR) of achieving remission in patients who have remote symptomatic aetiology
Arts et al 1999 | RR 0.66 (95%CI 0.46-0.95) MacDonald et al 2000 | RR 0.81 (95%CI 0.66-0.99)
Banu et al 2003 | OR 0.40 (95%CI 0.18-0.90) Hui et al 2007 | OR 0.10 (95%CI 0.05-0.25)
Box 8 | Odds ratio (OR) of achieving remission in patients with intellectual disability
Box 9 | Relative risk (RR) of achieving remission in patients as a function of Log N of Seizures before index
88
Table 12 | Studies predicting remission on or off AED (the DSEC Cohort)
Age (Years)*
Study
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure (% with Outcome)•
Independent Predictors
Risk Estimate (95%CI)‡
0-16 (6.7±4.5)
Arts, 1999 Prospective Hospital N=466
Y Y Y Y Y Only intake variables Included 1 status epilepticus in epilepsy definition
2yrs (Minimum)
6 months seizure-free immediately before 2yrs follow up (68.7%)
Log N of Seizures before intake Simple Partial vs. GTCS Infantile Spasms, Myoclonic, Atonic vs. GTCS R. Symptomatic vs. Idiopathic Cryptogenic vs. Idiopathic
OR: 1.51 (1.05, 2.19) OR: 3.15 (1.32,7.51) OR: 3.01 (1.45, 6.23) OR: 1.90 (1.06, 3.39) OR: 2.20 (1.25, 3.87)
0-16 (6.7±4.5)
Arts, 1999 Prospective Hospital N=466
Y Y Y Y Y Intake and 6 month variables Included 1 status epilepticus in epilepsy definition
2yrs (Minimum)
6 months seizure free immediately before 2yrs follow up (68.7%)
Simple Partial vs. GTCS Cryptogenic vs. Idiopathic Abnormal EEG at 6 months During 1
st 6 months after intake:
3 months seizure free >25 Seizures Log N seizures
OR: 2.72 (1.07, 6.89) OR: 1.95 (1.05, 3.61) OR: 2.21 (1.12, 4.36) OR: 0.32 (0.18, 0.58) OR: 2.20 (1.06, 4.56) OR: 1.99 (1.39, 2.85)
0-16 (6.7±4.5)
Arts 2004 Prospective Hospital N=453
Y
Y Y Y Y Only intake variables Included 1 status epilepticus in epilepsy definition
5yrs (Minimum)
1 year seizure free immediately before 5yrs follow up (76%)
Age <6 years at intake Not Idiopathic Aetiology + No history of febrile seizures No history of febrile seizures + Idiopathic Aetiology
OR: 0.62 (0.39, 0.99) OR: 3.72 (2.20, 6.30) OR 4.37 (1.70, 11.26)
0-16 (6.7±4.5)
Arts, 2004 Prospective Hospital N=453
Y Y Y Y Y Intake and 6 month variables Included 1 status epilepticus in epilepsy definition
5yrs (Minimum)
1 year seizure free immediately before 5yrs follow up (76%)
Female Post-ictal Signs Not Idiopathic Aetiology + No history of febrile seizures No history of febrile seizures + Idiopathic Aetiology ≤2 Seizure free months immediately before 6 month follow up
OR: 1.64 (1.00, 2.70) OR:2.23 (1.08, 4.63) OR:3.58 (2.05, 6.27) OR:5.28 (1.92, 14.51) OR: 4.47 (2.00, 9.99)
*Age range in years (Mean ± Standard Deviation); •The proportion reported is of those who achieved remission on or off AED; ‡Risk estimates for not achieving remission are presented; CI-Confidence Interval, Y-Yes, OR-Odds Ratio, GTCS-Generalised Tonic Clonic Seizures, EEG-Electroencephalogram, NS-Not Stated
89
Table 13 | Studies predicting remission on or off AED (Others 1)
Age (Years)*
Study
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure (% with Outcome)•
Independent Predictors
Risk Estimate (95%CI)‡
0-15 Banu, 2003(93)
Retrospective Hospital N=151
Y U N Y Y 1yr (Minimum)
3 months seizure free immediately before last follow up (49.7%)
Multiple Seizure Types Intellectual disability Abnormal EEG
OR: 4.42 (2.07, 9.57) OR: 2.49 (1.11, 5.70) OR:4.09 (1.53, 12.11)
0-15 Berg, 2001b Prospective Population N=594
Y Y Y Y Y 2-8yrs 2 year seizure free period during follow up (74%)
Log Initial Seizure Frequency Family History of Epilepsy Remote Symptomatic Aetiology Abnormal EEG Idiopathic Generalized Epilepsy Age 5–9 yrs at onset
HR: 0.90 (0.86, 0.95) HR: 0.58 (0.42, 0.81) HR: 0.63 (0.47, 0.84) HR: 0.71 (0.54, 0.93) HR: 1.65 (1.28, 2.12) HR: 1.23 (1.01, 1.50)
0-15 (4.8±1.6)
Sillanpaa, 1990
Mixed Population N=182
Y U Y Y Y Three seizures diagnosed epilepsy
23-39yrs 3yrs seizure free immediately before last follow up (76.4%)
High initial Seizure Frequency Occurrence of Status Epilepticus Abnormal Fine Motor Status
OR: 8.5 (3.1, 25.8) OR: 3.2 (1.5, 6.9) OR: 9.6 (3.4, 27.1)
0-15 (4.8±1.6)
Sillanpaa, 1993
Mixed Population N=178
Y U Y Y Y Three seizures diagnosed epilepsy
23-39yrs 12 seizure free months during last 10 years of follow up (77.5%)
Remote Symptomatic Aetiology High Initial Seizure Frequency Occurrence of Status Epilepticus No Seizure Freedom within 3 months of AED
OR: 2.9 (1.1, 8.2) OR: 4.6 (1.1, 19.3) OR: 11.4 (3.2, 41.0) OR: 3.6 (1.2-10.4)
0-15 (4.3)
Sillanpaa, 1998
Mixed Population N=176
Y U Y Y Y Three seizures diagnosed epilepsy
23-39yrs 5yrs seizure free immediately before last follow up (64%)
75-100% reduction in seizures within 3months of AED Complex Partial Seizures Atonic Seizures
HR: 0.27 (0.15, 0.49) HR: 3.33 (2.04, 5.56) HR: 4.55 (1.37, 14.3)
0-15 Sillanpaa, 2009
Prospective Population N=102
Y Y Y Y Y Three seizures diagnosed epilepsy Only prospectively identified patients St. Epilepticus excluded
11-40yrs 1yr seizure free immediately before last follow up (76%)
< Weekly seizures: Before AED During first year of AED
HR: 0.37 (0.2, 0.67) HR: 0.59 (0.35, 0.96)
0 – 19 Shinnar, 2000
Prospective Hospital N=182
Y Y Y Y Y Excellent identification. 10 recurrences defined refractoriness.
8.4yrs (Mean)
Not up to 10th seizure (71.4%)
Remote Symptomatic Aetiology 2nd Seizure within 1 year
HR: 2.13 (1.22, 3.71) HR: 6.94 (1.56, 30.7)
*Age range in years (Mean ± Standard Deviation); •The proportion reported is of those who achieved remission on or off AED; ‡Odds Ratio for not achieving remission are presented; CI-Confidence Interval, Y-Yes, U-Unsure, OR-Odds Ratio, HR-Hazard Ratio, AED-Antiepileptic Drug, EEG-Electroencephalogram, NS-Not Stated
90
Table 14 | Studies predicting remission on or off AED (Others 2)
Age (Years)*
Study
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure (% with Outcome)•
Independent Predictors
Risk Estimate (95%CI)‡
2 – 81 (19) CGSE, 1992 Mixed Hospital N=280
Y Y Y Y Y 0.1-7.25yrs 2 year seizure free period by 5 years follow up (67.5%)
≥2 Seizures before AED Mixed Seizure Types
HR: 1.49 (0.98-3.13) HR: 1.43 (0.96-2.70)
2 – 81 (19) CGSE, 1992 Mixed Hospital N=280
Y Y Y Y Y 0.1-7.25yrs 3yrs seizure free period by 5 years follow up (51.1%)
≥2 Seizures before AED Mixed Seizure Types
HR: 1.52 (0.94, 5.0) HR: 1.61 (0.98-4.35)
2 – 68 (24) Lohani, 2010 Retrospective Hospital N=284
Y U Y Y Y Not a newly diagnosed epilepsy cohort. 50% had seizures for >1yr before AED and only6.5% <1mo
6mo on AED (Minimum)
6mo seizure free immediately before last follow up (73.2%)
Neurological Deficit Failure of first AED
OR:6.69 (1.37, 32.67) OR:71.82 (16.51, 312.44)
5->50 MacDonald, 2000
Prospective Population N=289
Y Y Y Y Y Age < 5 excluded from analysis to avoid interactions
5.6-7.6yrs (IQR)
1 year seizure free period during follow up (NS)
Log N Seizures before index Log N Seizures (index - 6months) >10 Seizures before index
HR: 0.81 (0.66, 0.99) HR:0.59 (0.50, 0.70) HR: 1.89 (1.14, 3.13)
5->50 MacDonald, 2000
Prospective Population N=289
Y Y Y Y Y Age < 5 excluded from analysis to avoid interactions
5.6-7.6yrs (IQR)
5 years seizure free period during follow up (NS)
Log N Seizures (index - 6months) >10 Seizures before index
HR: 0.64 (0.51, 0.81) HR: 1.80 (1.10, 2.92)
All ages (NS)
Shafer, 1988 Retrospective Population N=298
Y U N Y Y Patients without EEG or information on GTC Seizure excluded
≈13-17yrs Mean (NS)
5 years seizure free period during follow up (NS)
Unknown Aetiology No Gen. Spike Waves on 1st EEG No Gen. Tonic Clonic Seizures
HR: 2.27 (NS) HR: 1.58 (NS) HR: 1.4 (NS)
12-85 (28.7±12.4)
Abduljabbar, 1998
Mixed Hospital N=826
Y Y Y Y Y 1yr (Minimum)
1 year seizure free period during follow up (80%)
≥2 AEDs Compliance Therapeutic Drug Level Short duration btw onset & AED
OR: 7.39 (NS) OR: 40.44 (NS) OR: 3.0004 (NS) OR: 1.003 (NS)
15-79 (34) Hui, 2007 Mixed Hospital N=260
Y Y Y Y Y 1-17yrs 7years (Mean)
1yr seizure free immediately before last follow up (60%)
Intellectual disability Mesial Temporal Sclerosis
OR:9.39 (3.98, 22.12) OR: 7.6 (3.53, 16.4)
18-67 (44±12)
Lossius, 1999
Retrospective Population N=669
Y Y Y Y Y NS 1yr seizure free immediately before last follow up (72.5%)
≥2 AEDs vs. No Treatment Age ≥ 50 years
OR: 5.6 (2.7, 11.9) OR: 1.7 (1.1, 2.6)
*Age range in years (Mean ± Standard Deviation); •The proportion reported is of those who achieved remission on or off AED; ‡Odds Ratio for not achieving remission are presented; CI-Confidence Interval, Y-Yes, U-Unsure, N-No, OR-Odds Ratio, HR-Hazard Ratio, AED-Antiepileptic Drug, EEG-Electroencephalogram, NS-Not Stated
91
Table 15 | Consistent predictors and non-predictors of remission on or off AED (1)
Art
s, 1
99
9
Art
s, 1
99
96
mo
Art
s, 2
00
4
Art
s, 2
00
46
mo
Ban
u, 2
00
3
Be
rg, 2
00
1b
Silla
np
aa, 1
99
0
Silla
np
aa, 1
99
3
Silla
np
aa, 1
99
8
Silla
np
aa, 2
00
9
Shin
nar
, 20
00
CG
SE, 1
99
2 2
CG
SE, 1
99
23
Loh
ani,
20
10
Mac
Do
nal
d, 2
00
01
Mac
Do
nal
d, 2
00
03
Shaf
er,
19
88
Ab
du
ljab
bar
, 19
98
Hu
i, 2
00
7
Loss
ius,
19
99
Demographics
Age at Onset ✘ ✘ Θ ✘ Θ Θ ✘ ✘ ✘ ✘
Gender ✘ ✘ ✔* ✘ Θ Θ ✘ ✘ ✘ ✘ ✘ ✘
AED Therapy
≥2 AEDs ✔ ✔
Epilepsy Before AED, Intake or Index
Duration ✔ ✘ ✘ ✘ ✔
N of Seizures before Intake ✘ Θ ✔ ✔
Log N of Seizures before Intake ✔ ✔ ✔
Seizure Frequency ✔ ✔ ✔ ✔ Θ ✔ ✔ ✔
Early Epilepsy Characteristics
Log N (+1) of Seizures from 0 - 6mo ✔ ✔ ✔
Status Epilepticus (SE) ✔ ✔ ✔ Θ
EEG
Intake/Early - EEG Abnormal ✘ ✘ ✔ ✘ ✘ ✘ ✘ ✘
Intake Epileptiform Abnormality Θ ✘ ✘
Gen (Epileptiform) Spike and Wave ✘ ✔
Focal ✘ ✘
✔ - Variable is significant or retained in multivariate model, ✔ - Variable is only significant on univariate analysis, ✘- Variable is not significant on univariate analysis Θ – Variable is not reported in univariate analysis, but reported as not significant on multivariate analysis *In this study, female was the gender retained in the model
92
Table 16 | Consistent predictors and non-predictors of remission on or off AED (2)
Art
s, 1
99
9
Art
s, 1
99
96m
o
Art
s, 2
00
4
Art
s, 2
00
46m
o
Ban
u, 2
00
3
Be
rg, 2
00
1b
Silla
np
aa, 1
99
0
Silla
np
aa, 1
99
3
Silla
np
aa, 1
99
8
Silla
np
aa, 2
00
9
Shin
nar
, 20
00
CG
SE, 1
99
2 2
CG
SE, 1
99
23
Loh
ani,
20
10
Mac
Do
nal
d, 2
00
01
Mac
Do
nal
d, 2
00
03
Shaf
er,
19
88
Ab
du
ljab
bar
, 19
98
Hu
i, 2
00
7
Loss
ius,
19
99
Seizure Type
Mixed Seizure Types ✔ ✔ ✔
Seizure Type ✘ ✘ ✘
Generalised Onset ✘ ✘ ✘
Partial ✘ ✘ ✘
Complex Partial ✘ ✔
Secondarily Generalised ✘ ✘ ✘ ✘
Atonic ✘ ✔
Generalised Tonic Clonic ✘ ✘ ✘ ✘ ✔
Aetiology | Syndrome
Remote Symptomatic Aetiology ✔ ✔ ✔ ✔ ✘ ✘ ✘ ✘ ✔ ✔ ✔
Cryptogenic ✘ ✘ ✔
West Syndrome ✔ ✔
Cognition
Abnormal Cognitive Development ✔ ✔ ✔ Θ ✔
Neurological Sign
Neurological Examination ✔ ✘ Θ ✘ ✘ ✔ ✔ ✔
Neuroimaging Findings ✔ ✘ ✘ ✔ ✘
Others
Family History ✘ ✘ ✔ ✘ ✘ ✘ ✘
Perinatal Asphyxia ✘ ✘ ✘
Neonatal Seizures ✘ ✘ ✘
Febrile Seizure ✘ ✘ ✘ ✘ ✘ ✘
Tumour ✘ ✘ ✘
Vascular Malformation ✘ ✘ ✘
✔ - Variable is significant or retained in multivariate model, ✔ - Variable is only significant on univariate analysis, ✘- Variable is not significant on univariate analysis Θ – Variable is not reported in univariate analysis, but reported as not significant on multivariate analysis
93
The natural logarithm transformation of the number of seizures from intake to 6 months was
also consistently identified as independent predictor of remission in Arts et al 1999(65) (in the
model including 6 months variables) and MacDonald et al(101) (although it is also retained in
the model predicting a 5 year seizure-free period, the model predicting I year free of seizures
during follow up is used in this comparison owing to its greater similarity to the outcome
measure in Arts et al 1999):
No demographic variable was found to consistently predict remission; only 1 study,(66) found
gender to be an independent predictor on multivariate analysis, albeit with borderline statistical
significance. Other variables not associated with remission include age at onset, abnormal EEG
at intake and the range of seizure types considered, as were factors like family history of
epilepsy, history of neonatal seizures and the specific aetiological factors considered.
The 2 models derived in Arts et al 1999(65) were externally validated in a temporal cohort from
the same centre as the derivation cohort. (Table 17) The model derived from only variables
assessed at intake showed poor discriminative ability (AUC, <0.70) in both internal and external
validation. However, the model that was fitted with both intake and 6 month variables showed
good discrimination (AUC, >0.70), with the AUC reducing from 0.78 in internal validation to 0.71
when externally validated. The proportion of patients with correct prediction (accuracy) also
reduced from 73% to 63% in external validation.
The 2 models were calibrated, with a plot of 4 risk groups of unequal size, which suggest that
the percentage of children correctly predicted not to achieve remission did increase with the
predicted chance of not achieving remission. The calibration was however not assessed using a
formal statistical test.
Arts et al 1999 | RR 0.50 (95%CI 0.35-0.72) MacDonald et al 2000 | RR 0.59 (95%CI 0.50-0.70)
Box 10 | Relative risk of achieving remission in patients as a function of Log N of Seizures from index to 6 months
94
Table 17 | Externally validated models predicting remission on or off antiepileptic drugs
Predictors in Model
Derivation Cohort
Internal Validation: Performance of Model in Derivation Cohort
Validation Study
Validation Cohort
External Validation: Performance of Model in Validation Cohort Sc
ore
Easy
to
Use
Imp
ort
ant
Var
iab
le
Pro
bab
ility
or
Act
ion
Clin
ical
Use
Stu
die
d
Log N of Seizure before intake >/<25 Seizures before intake Seizure Type -Generalised Tonic Clonic -Complex Partial -Simple Partial -Absences -Other Types Aetiology -Remote symptomatic -Cryptogenic, -Idiopathic Yes/No Neurological signs
Arts 1999 Only intake variables Included 1 status epilepticus in epilepsy definition Remission defined by 6 months seizure free immediately before 2yrs follow up Pre-Test Probability (31%)
AUC 0.69 (0.64, 0.74) Best probability cut-off (38%) Sensitivity (61.6%) Specificity (69.1%) PPV (47.2%) NPV (20.0%) Correct Prediction (66.7%)
Geerts 2006
Temporal Validation N=273 2 years (Minimum Follow up) The 2 cohorts were similar except there were more children with normal EEG findings than in the derivation cohort. Pre-Test Probability (34%)
AUC 0.62 (0.55, 0.69) Probability cut-off (38%) Sensitivity (60.0%) Specificity (61.4%) PPV (44.5%) NPV (25.1%) Correct Prediction (59.9%)
Y N Y Pr N
Seizure Type -Generalised Tonic Clonic -Complex Partial -Simple Partial -Absences -Other Types Aetiology -Remote symptomatic -Cryptogenic, -Idiopathic During 1
st 6 mo after intake:
-Log N Seizures ->/<25 Seizures -Yes/No 3 month remission EEG at Intake -Normal, -Epileptiform,-Other EEG 6 mo after intake -Normal,-Epileptiform, -Other
Arts 1999 Intake and 6 month variables Included 1 status epilepticus in epilepsy definition Remission defined by 6 months seizure free immediately before 2yrs follow up Pre-Test Probability (31%)
AUC 0.78 (0.73, 0.82) Best probability cut-off (34%) Sensitivity (72.6%) Specificity (73.1%) PPV (54.8%) NPV (14.4%) Correct Prediction (73.0%)
Geerts 2006
Temporal Validation N=273 2 years (Minimum Follow up) The 2 cohorts were similar except there were more children with normal EEG findings than in the derivation cohort. Pre-Test Probability (34%)
AUC 0.71 (0.64, 0.78) Probability cut-off (34%) Sensitivity (67.4%) Specificity (60.2%) PPV (46.6%) NPV (21.8%) Correct Prediction (62.7%)
Y N Y Pr N
AUC-Area Under Receiver Operator Characteristics (ROC) Curve, PPV-Positive Predictive Value, NPV-Negative Predictive Value, NS-Not Stated, Pr-Probability, N-No, Y-Yes, AED-Antiepileptic Drug
95
PREDICTING INTRACTABILITY (TYPE IV) Twelve studies, all conducted in childhood epilepsy cohorts, attempted to identify independent
predictors of medically intractable seizures. Six of the analyses were conducted on the entire
cohort, (Table 18) while the other 6 fitted their model with a case control analysis nested within
cohorts. (Table 19 and 20)
The proportion of patients with intractable seizures in the cohorts varied among the studies
depending on design and how medical intractability was defined and conceptualised in each
study. Most of the studies defined intractability as having “at least 1” or “more than 1” seizure
per month for about 1 year (ranged between at least 6 months and at least 2 years) after more
than 2 antiepileptic drug trials singly or in combination, at the maximum tolerable dose.
The frequency of seizures that define medical intractability in each study influenced the
proportion of patients that met the criteria. The studies that defined intractability as having “1
or more” seizures per month(72, 75, 76) had more patients who met the criteria for
intractability (13% to 14%) compared to the studies(73, 74) that defined intractability as only
“more than 1” seizure per month (9% to 10%). However, in a retrospective, hospital based study
reported by Berg et al 1996(78), with the seizure frequency that defined epilepsy being “1 or
more” seizures per month, one third (32.5%) of the patients met the criteria for medical
intractability.
Two of the studies reported by Oskoui et al(72)and the report by Ko et al(79) included patients
who required epilepsy surgery during follow up as medically intractable even when they did not
meet the usual criteria for intractability. The same 2 studies by Oskoui et al(72) also included
patients who required the use of the ketogenic diet to control seizures, and only patients who
have had recurrent seizures in the last 6 months of follow up in a category designated as having
“poor outcome.” 7% of the cohort was included in this category.
The age of children at the onset of seizures, kept as a continuous variable, was retained in the
models predicting medical intractability in three different cohorts, with nested case control
analysis (75, 78, 79):
Casetta et al 1999 | RR 0.98 (95%CI 0.97-1.28) Ko et al 1999 | RR 0.86 (95%CI 0.75-0.97) Berg et al 1996 | RR 0.78 (95%CI 0.69-0.89)
Box 11 | Relative risk (RR) of having intractable seizures for each increasing year of age at onset of seizures
96
Table 18 | Studies predicting intractability (full cohort)
Age (Years)*
Study
Design Setting St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure (% with Outcome)
Independent Predictors
Risk Estimate (95%CI)
0-15 Berg, 2001a Prospective Population
Y Y Y Y N 5yrs (Median)
>1 Seizure per month for ≥1.5yrs with >2AEDs (10%)
Cryptogenic and Symptomatic Generalized Syndrome Idiopathic Syndrome Log Initial Seizure Frequency Abnormal EEG (Focal Slowing) Age 5 to 9 at Onset Provoked, Non-Febrile and Neonatal Status Epilepticus Myoclonic Seizures + Log Initial Seizure Frequency
HR: 3.01 (1.32, 6.85) HR: 0.23 (0.09, 0.63) HR: 1.41 (1.23, 1.62) HR: 2.31 (1.13, 4.74) HR: 0.42 (0.19, 0.92) HR: 5.96 (2.00, 17.71) HR: 0.73 (0.57, 0.95)
0-14 (4.8±3.8)
Ramos-Lizana, 2009
Mixed Hospital
Y Y Y Y Y All variables measured at 6 months of follow up
2-11.6yrs >1 Seizure per month for ≥1.5yrs with >2AEDs (8.7%)
Age <1 year at Onset Idiopathic Aetiology >1 Seizure (diagnosis -6 months)
HR: 2.6 (1.0, 6.9) HR: 0.2 (0.0, 0.8) HR: 4.8 (1.8, 13.0)
2-17 (7.6±3.7)
Oskoui, 2005 Retrospective Hospital
Y P Y Y Y Only onset variables Data collection was remarkably good
2-13.6yrs ≥1 Seizure per month for >1yr with >2AEDs (12.8%)
Multiple Seizure Types OR: 17.4 (4.8–63.1)
2-17 (7.6±3.7)
Oskoui, 2005 Retrospective Hospital
Y P Y Y N Onset and 1yr variables Data collection was remarkably good
2-13.6yrs ≥1 Seizure per month for >1yr with >2AEDs (12.8%)
Multiple Seizure Types Intellectual disability Seizure 6 to 12 months on AED
OR: 6.5(1.9–35.4) OR: 7.2 (1.0–50.8) OR: 70.4 (7.5–661.4)
2-17 (7.6±3.7)
Oskoui, 2005 Retrospective Hospital
Y P Y Y N Only onset variables Data collection was remarkably good Outcome included having required epilepsy surgery or ketogenic diet
2-13.6yrs Recurrent seizures on adequate AED 6 months immediately before last follow up (6.9%)
Multiple Seizure Types Intellectual disability Idiopathic Epilepsy
OR: 14.7 (4.7–46.1) OR: 3.3 (1.1–10.2) OR: 0.13 (0.03–0.52)
2-17 (7.6±3.7)
Oskoui, 2005 Retrospective Hospital
Y P Y Y N Onset and 1yr variables Data collection was remarkably good Outcome included having required epilepsy surgery or ketogenic diet
2-13.6yrs Recurrent seizures on adequate AED 6 months immediately before last follow up (6.9%)
Multiple Seizure Types Intellectual disability Seizure 6 to 12 months on AED
OR: 8.9 (2.6–31.2) OR: 8.9 (2.4–32.7) OR: 21.6 (6.3–74.0)
*Age range in years (Mean ± Standard Deviation); CI-Confidence Interval, Y-Yes, P-partly, N-No, HR-Hazard Ratio, OR-Odds Ratio, AED-Antiepileptic Drug, EEG-Electroencephalogram
97
Table 19 | Studies predicting intractability (nested case control)
Age (Years)*
Study
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure
Independent Predictors
Risk Estimate (95%CI)
0-19 Cases 3.3 Controls 6.4
Casetta, 1999
Prospective Population N=222 Cases 31 Controls 95
Y Y Y Y Y Two models – I with age as continuous variable and II with age dichotomous 13.96% of patients in the cohort had intractable seizures
Cases 20.97yrs Controls 20.98yrs
Cases ≥1 Seizure per month for ≥2yrs with >2AEDs Controls 5 years seizure free period during follow up
Model I Age at Onset (in years) Remote Symptomatic Aetiology Weekly Seizures before AED Model II Age <1 year at Onset Remote Symptomatic Aetiology Weekly Seizures before AED
OR: 0.98 (0.97, 1.28) OR: 8.6 (2.3, 32.2) OR: 4.5 (1.3, 16) OR: 3.9 (1.1, 14.25) OR: 8.6 (2.4, 26.1) OR: 4.1 (1.3-13.8)
0-15 Kwong, 2003
Mixed Population N=309 Cases 44 Controls 211
Y Y Y Y Y 14.24% of patients in the cohort had intractable seizures
3yrs (Minimum) Cases 5.88yrs Controls 4.67yrs
Cases ≥1 Seizure per month for ≥2yrs with ≥3AEDs Controls 2 year seizure free period during follow up
Daily Seizures before AED ≥3 Seizures 6-12 months on AED History of Febrile Seizure Abnormal Neurodevelopmental Status (Intellectual disability and/or Cerebral Palsy)
OR: 1.58 (1.0, 2.51) OR:21.86(7.35,65.01) OR:4.83 (1.31, 17.83) OR:18.16(5.19,63.61)
Cases 1.58±2.38 Controls 5.83±2.54
Chawla, 2002
Mixed Hospital N=(NS) Cases 50 Controls 50
U U N U Y Equal number of cases and controls were selected from the cohort The total number of base population not presented
1yr (Minimum)
Cases ≥1 Seizure per month for ≥0.5yr with ≥2AEDs Controls 6 months seizure free immediately before last follow up
Age <1 year at Onset Remote Symptomatic Aetiology Neurological Impairment Initial Seizure Type (Myoclonic or Infantile Spasms)
OR: 11.7 (2.95,46.43) OR: 2.91 (1.14, 7.44) OR12.25 (3.58,41.89) OR: 10.36 (2.39, 44.94)
Cases 1.8 Controls 5.8
Berg, 1996
Retrospective Hospital N=234 Cases 76 Controls 96
N Y Y Y Y Case control study nested in a cohort 32.5% of patients in the cohort had intractable seizures
Cases 6yrs Controls 5yrs
Cases ≥1 Seizure per month for ≥2yrs with >2AEDs Controls 2 year seizure free period during follow up
Age at Onset (in years) Remote Symptomatic Epilepsy Occurrence of Status Epilepticus Infantile Spasms
OR: 0.78 (0.69, 0.89) OR: 2.24, (1.05, 4.80) OR: 3.30 (1.06,10.28) OR: 10.42 (1.27, 85.39)
*Age range in years (Mean ± Standard Deviation); CI-Confidence Interval, Y-Yes, N-No, U-Unclear, OR-Odds Ratio, AED-Antiepileptic Drug, NS-Not Stated
98
Table 20 | Studies predicting intractability (nested case control by Ko et al)
Age (Years)*
Study
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure
Independent Predictors
Risk Estimate (95%CI)
0-18 Cases 2.9 Controls 5.2
Ko, 1999
Retrospective Hospital N=(NS) Cases 144 Controls 39
N Y Y U N Patients without EEG at intake were excluded Only clinical variables entered in this model The total number of base population not presented
2yrs (Minimum) Cases 5.3yrs Controls 4.9yrs
Cases “Continued Seizures” with ≥3AEDs or having required epilepsy surgery Controls 1 year seizure free period during follow up
Age at Onset (in years) Remote Symptomatic Aetiology Occurrence of Status Epilepticus Simple Partial Seizures Absence Seizures Tonic Seizures
OR: 0.86 (0.75, 0.97 ) OR: 4.41 (1.68, 13.0) OR: 7.7(1.42, 143.48) OR: 5.11(1.21, 36.08) OR: 0.92 (0.32, 2.70) OR:13.5(2.56,249.74)
0-18 Cases 2.9 Controls 5.2
Ko, 1999
Retrospective Hospital N=(NS) Cases 144 Controls 39
N Y Y U N Patients without EEG at intake were excluded Only EEG variables entered in this model The total number of base population not presented
2yrs (Minimum) Cases 5.3yrs Controls 4.9yrs
Cases “Continued Seizures” with ≥3AEDs or having required epilepsy surgery Controls 1 year seizure free period during follow up
Photo-convulsive Response 3 Hz Spike and Wave Diffuse Slowing Focal Spike and Wave Frequent Sharp Wave/ Spike (>1/60 s)
OR: 0.10 (0.02, 0.51) OR: 0.21 (0.05, 0.74) OR: 2.59 (1.05, 6.84) OR: 5.21 (1.25,38.16) OR: 1.68 (0.67, 4.43)
*Age range in years (Mean ± Standard Deviation); CI-Confidence Interval, Y-Yes, P-partly, N-No, HR-Hazard Ratio, OR-Odds Ratio, AED-Antiepileptic Drug, NS-Not Stated
99
Table 21 | Consistent predictors and non-predictors of intractability
Be
rg, 2
00
1a
Ram
os-
Liza
na,
20
09
Osk
ou
i, 2
00
5in
trac
tab
le
Osk
ou
i, 2
00
5in
trac
tab
le*
Osk
ou
i, 2
00
5p
oo
r
Osk
ou
i, 2
00
5p
oo
r *
Cas
ett
a, 1
99
9
Kw
on
g, 2
00
3
Ch
awla
, 20
02
Be
rg, 1
99
6
Ko
, 19
99
CLI
Ko
, 19
99
EEG
Demographics
Age at Onset (in years)* ✔ Θ Θ Θ Θ ✔ ✔ ✔ ✔
Age at onset <1 year† ✔ ✔ ✔ ✔
Gender ✘ ✘ Θ Θ Θ Θ ✘
Epilepsy Before AED, Intake or Index
Seizure Frequency - Daily ✔ ✔ ✔ ✔
Early Epilepsy Characteristics
Status Epilepticus (SE) ✔ ✘ ✔ ✔ ✘ ✔ ✘
EEG
Intake/Early - EEG Abnormal ✘ ✘ Θ Θ Θ Θ ✔ ✘
Focal Slowing ✔ ✘
Seizure Type
Mixed Seizure Types ✔ ✘ ✔ ✔ ✔ ✔
Generalised Onset ✘ ✘
Partial ✘ ✔ ✘
Simple ✘ ✔
Complex ✘ ✘
Myoclonic ✔ ✔
Absence ✔ ✔ ✔
Generalised Tonic Clonic ✘ ✘
Aetiology | Syndrome
Remote Symptomatic Aetiology ✔ ✔ Θ Θ Θ Θ ✔ ✔ ✔ ✔ ✔
Idiopathic ✔ ✔ Θ Θ ✔ Θ
West Syndrome ✔ ✔ ✔
ILAE Syndrome ✔ Θ Θ Θ Θ ✔
Cognition
Abnormal Cognitive Development ✔ Θ ✔ ✔ ✔ ✔ ✔
Neurological Sign
Neurological Examination Θ Θ Θ Θ ✔
Normal Neuroimaging Findings ✔ ✔ ✔ ✔
Others
Motor disability ✘ ✔
Family History ✘ Θ Θ Θ Θ ✘ ✘ ✘
Neonatal Seizures ✔ ✘ ✘ ✔ ✔ ✔ ✘
Febrile Seizure ✘ ✘ ✘ ✔ ✘ ✘ ✘
Multiple Seizures within 24 Hours ✔ ✔
Microcephaly ✔ ✔ ✘
✔ - Variable is significant or retained in multivariate model, ✔ - only significant on univariate analysis, ✘- not significant on univariate analysis Θ – not reported in univariate analysis, but reported as not significant on multivariate analysis
*Where “age in years” was retained, the risk of intractability reduces with each additional year increase in the age of onset.
†“Age at onset < 1 year” was retained as a variable which increases the risk of intractability.
100
Three studies also retained onset of seizures in infancy (age at onset < 1year) as an independent
predictor of medically intractable seizures (74, 75, 94):
The presence of remote symptomatic aetiology was also found to be a consistent predictor of a
patient developing medically intractable seizures in 4 studies, all nested cohort studies, with the
risk estimate from Casetta et al(75) being remarkably higher than the rest:
Three studies also retained idiopathic aetiology/syndrome as an independent predictor of
patients developing medically intractable seizures, with similar estimates although the study
from Oskoui et al (72) was one in which “intractability” defined as “poor outcome” (patients
who required epilepsy surgery and/or the use of ketogenic diet to control seizures, or who have
had recurrent seizures in the last 6 months of follow up) was rather atypical:
Intellectual disability was another consistent predictor of patients developing medically
intractable seizures, with the variable retained in three of the 4 studies on intractability
reported by Oskoui et al(72) and in the only study reported by Kwong et al(76). The result of the
study from Oskoui et al(72) whose definition of medical intractability is most similar to that from
Kwong et al(76) is presented for comparison. Kwong et al(76) includes patients with cerebral
palsy, and the compared estimate from Oskoui et al(72) was assessed alongside variables
collected at 1 year follow up.
However, as Oskoui et al(72) did not provide enough data (absolute risk of intractability in non-
intellectually disabled patients) to compute the relative risk, the estimates are compared as
odds ratio:
Ramos-Lizana et al 2009 | RR 0.20 (95%CI 0.0-0.80) Oskoui et al 2005 | RR 0.13 (95%CI 0.03-0.52) Berg et al 2001a | RR 0.23 (95%CI 0.09-0.63)
Oskoui et al 2005 | OR 7.2 (95%CI 1.0-50.8) Kwong et al 2003 | OR 18.2 (95%CI 5.2-63.6)
Box 14 | Relative risk (RR) of having intractable seizures in patients with idiopathic aetiology
Box 15 | Odds ratio (OR) of having intractable seizures in patients with intellectual disability
Chawla et al 2002 | RR 2.06 (95%CI 1.11-3.10) Casetta et al 1999 | RR 5.48 (95%CI 2.10-9.64) Ko et al 1999 | RR 1.34 (95%CI 1.15-1.43) Berg et al 1996 | RR 1.67 (95%CI 1.04-2.36)
Box 13 | Relative risk (RR) of having intractable seizures in patients with remote symptomatic aetiology
Ramos-Lizana et al 2009 | RR 2.6 (95%CI 1.0-6.9) Chawla et al 2002 | RR 3.1 (95%CI 2.0-3.6) Casetta et al 1999 | RR 2.6 (95%CI 1.1-4.3)
Box 12 | Relative risk (RR) of having intractable seizures with onset of seizures in infancy (age <1 year)
101
Having mixed seizure types was only found to be a predictor of intractability in studies reported
by Oskoui et al(72). Other potential predictors related to seizure type were not confirmed to be
consistently predictive of intractability, as were the occurrence of status epilepticus, family
history of epilepsy or history of neonatal seizures, seizure frequency, and abnormal EEG.
None of the models were externally validated.
PREDICTING NO REMISSION AFTER RELAPSE (TYPE V) Only 1 study, Sillanpaa and Schmidt 2006(25) investigated the predictors of not achieving
remission after 1 or more relapses following initial remission. In the study, 14% of the cohort
belonged in this category. The only independent predictor was having remote symptomatic
aetiology with the odds 8 [8.2 (95%CI 2.7-24.8)] times as high compared to those without
remote symptomatic aetiology. Intellectual disability, localisation related epilepsy and temporal
lobe epilepsy were not included in the model. The model was not externally validated.
102
Table 22 | Studies predicting no remission after relapse
Age (Years)*
Study
Design Setting N St
ud
y P
arti
cip
atio
n
Stu
dy
Att
riti
on
Pro
gno
stic
Fa
cto
r
Ou
tco
me
An
alys
is
Remarks
Follow Up
Outcome Measure
Independent Predictors
Risk Estimate (95%CI)
0-15 Sillanpaa, 2009
Prospective Population N=144 Cases 20 Controls 97
Y Y Y Y Y Three seizures diagnosed epilepsy Cases and controls nested within a selection of only prospectively identified patients in Turku cohort
11-42yrs Cases Patients who after relapse, did not achieve 5 years seizure free status immediately before last follow up Controls 5 years seizure free immediately before last follow up
Remote Symptomatic Aetiology OR: 8.2 (2.7–24.8)
*Age range in years (Mean ± Standard Deviation); CI-Confidence Interval, Y-Yes, HR-Hazard Ratio, OR-Odds Ratio
Table 23 | Predictors and non-predictors of no remission after relapse
Silla
np
aa, 2
00
6
Aetiology | Syndrome
Remote Symptomatic Aetiology ✔
Cognition
Abnormal Cognitive Development Θ
Others
Localisation Related Epilepsy Θ
Temporal Lobe Epilepsy Θ
✔ - Variable is significant or retained in multivariate model, Θ – not reported in univariate analysis, but reported as not significant on multivariate analysis
103
104
5 CHAPTER FIVE: DISCUSSSION This chapter presents a general discussion and reflection on the results of the review. The
chapter is presented according to recommendations that advocate structured discussion of the
results of scientific research.(38) The principal findings of 1) the literature search; 2) the quality
appraisal of studies and reporting characteristics of publications; 3) the study classification; and
4) the studies within each category, were stated and interpreted. Thereafter, possible
explanations for the results are presented and their implications for clinical practice are stated
where applicable. There are also suggestions for future studies and the direction of future
research for each set of results. The strengths of the study are highlighted, and the weaknesses
are also discussed also with a view to making recommendations for future studies.
5.1 THE LITERATURE SEARCH MEDLINE alone retrieved 91% of all eligible publications, with an overlap of 42% (14 out of 33
publications) between MEDLINE and EMBASE. This distribution of search results mirrors that of
the study by Royle et al(240) which sought to ascertain how many databases were necessary for
a comprehensive coverage of observational studies of diabetes published in English language.
MEDLINE alone accounted for 94% of all the articles retrieved and the rest were from EMBASE.
No additional English language publication was retrieved beyond these 2 databases. This
suggests that the search in this thesis was likely exhaustive; a notion further reinforced by the
fact that the screening of reference lists of eligible publications did not yield any eligible study.
These findings further confirm that the 2 databases are generally complimentary.(241, 285-288)
The 3 eligible publications unique to EMBASE(96, 97, 100) were all published in journals indexed
on MEDLINE.(289) That 2 of the 3 publications were published in 2010 may support the claim
that EMBASE is much more current than MEDLINE.(242) The fact that all the articles were
published in the last 15 (1988 to 2010) years may reflect the period when multivariate analysis
became popular due to proliferation of computer software packages that made the statistical
technique much easier to conduct, beginning in the late 1980s.(263)
One journal, Epilepsia, accounted for about a third of the 33 eligible publications. Future
systematic reviews of observational studies in epilepsy may therefore also benefit from hand-
searching Epilepsia. Evidence from the search and systematic reviews of intervention studies
suggest that hand searching of selected high impact journals may be necessary to ensure
comprehensive search as the processes associated with indexing and electronic search are not
infallible. (290, 291) The greatest value of hand searching has been found to be in identifying
publications in supplement editions and abstract sections of journals.(291) This is particularly
the case in MEDLINE search(291) and in this review, as 1 (292) of the 4 EMBASE unique
publications (although not eligible) was published in the abstract section of a supplement
edition of Epilepsia, a MEDLINE-indexed relatively high impact journal.
105
Notwithstanding that the care of patients with newly diagnosed epilepsy is often initiated and
followed up in primary care,(293, 294) only 1 out of the 33 publications was in a non-specialist
journal. This may have implications for the uptake of research findings, as the category of
physicians who often have first contact with patients may not be specialists and they may be the
physicians with the responsibility to assess prognosis and plan further care or referral.(295)
5.2 QUALITY AND REPORTING CHARACTERISTICS The report of important study items in the eligible publications was generally poor, which
impacted negatively on the ability to conduct a detailed critical appraisal of the quality of some
studies, this being so despite seeking out referenced publications for further information on the
studies.
No study gave a power calculation to justify their sample size. However, the more important
issue, which affects the stability and reliability of risk estimates from multivariate models is the
number of patients with the outcome under consideration in relation to the number of
independent variables in the model.(263, 296) Nine out of the 47 studies with multivariate
analysis failed to meet the criterion that there must be at least 10 times the number of patients
with the less frequent outcome per prognostic variable in the model. Two (66, 77) of the 9
studies had more than 9 events per variable. The overall small sample size may also be a major
source of unreliability, especially with the use of a stepwise algorithm for selecting the variables
to retain in the model, if continuous markers are dichotomized, (an act that effectively reduces
the sample size by 30% or more) and when interactions between variables is investigated.(45,
296)
Eight of 31 publications reported testing for interaction terms, 16 reported the use of stepwise
selection of variables and only 4 handled continuous variables appropriately. However, Altman
and Lyman suggested, rather arbitrarily, that studies that use stepwise algorithm, dichotomise
continuous variables and investigate interaction terms should be based on at least 250 to 500
patients with outcome. In this review, only 2 publications (reporting 4 studies) had such rate of
outcome. The 2 studies were reports of international collaborative studies. The sample size of
the cohorts included in this review ranged from 102 to 1443 (median 287, mean 390).
The practice in 2 previous systematic reviews of prognosis studies, as in this review, was to have
100 participants in a study as minimum requirement for inclusion.(234, 235) However, to ensure
robustness of results of multivariate analyses, future prognosis studies of seizure outcome in
newly diagnosed patients with epilepsy may benefit from international collaborations and the
meta-analysis or reanalysis of individual patient data from different cohorts. Having a large
sample size can only improve the precision and stability of models, but will not necessarily make
up for inherent weaknesses in study design and execution.(45, 296)Therefore, these
106
international collaboration studies will also ensure that other quality measures such as
prospective case ascertainment and long follow up to allow patients to be reliably grouped into
appropriate outcome categories.
Less than 60% of the publications reported the statistical package or software used to conduct
the multivariate analysis, a practice which Concato et al(263) described as “analogous to a
laboratory researcher [not] indicating the particular experimental protocol used for physiologic
measurements.” The reporting of issues directly related to multivariate analysis was also
deficient. The test for collinearity was reported in only 1 publication; model assumptions were
tested in 2, missing data was mentioned in 8, and the statistical criteria for including variables in
models were stated in only 50% of the 31 papers reporting studies with multivariable models.
However, all but 1 paper(68) reported a measure of follow up, while all but 2 (101, 104)
reported the proportion with outcome.
The poor reporting of studies is not unique to prognosis studies; it affects observational studies
in general.(297) Therefore in 2007, the STROBE (Strengthening the Reporting of Observational
Studies in Epidemiology) checklist was produced in order to serve as guidelines for reporting
observational studies.(298) The checklist consists of a flexible 22-item checklist which provides
reporting recommendations for all sections of manuscripts of the 3 types of observational study
design (cohort, case control and cross-sectional) while allowing authors to utilize their
preferences and creativity when selecting the order and format of presenting the details.(298)
Eighteen of the items are applicable to all the 3 study designs. Four are design-specific:
information in the methods section regarding participants (item 6) and statistical methods (item
12); information in the results section regarding descriptive data (item 14) and outcome data
(item 15).(40) The checklist is presented in Appendix V
However only 2 journals (Lancet Neurology and Neurology, containing 1 publication each) out of
the total 17 journals where the papers in the review were published have endorsed the STROBE
statement by way of referring to it in their Instructions for Authors.(299) The 2 articles in the
STROBE-endorsing journals were published before the STROBE statement was produced,
therefore it was not possible to compare the reporting characteristics of the publications with
the others published in journals that are yet to endorse the STROBE guidelines.
5.3 THE CLASSIFICATION OF STUDIES The task of developing a method for classification of studies to assist quality assessment and to
guide the present and also future systematic reviews and meta-analyses is an important role for
systematic reviews. This is especially important for observational studies, given the range of
differences in quality, methods and potential for heterogeneity.(300) This thesis developed a
classification system for studies of prognosis of seizure outcome in epilepsy. The classification is
based on: 1) the potential seizure outcome category where a newly diagnosed patient could
107
belong e.g. immediate remission, remission off medication, remission on or off medication or
medical intractability; and 2) how in the analysis to identify independent predictors of each
outcome, patients belonging to different outcome groups are compared to identify exposure or
epilepsy characteristics that are significantly different between the groups. The studies included
in this review were classified into 5 types and systematically reviewed within the categories.
5.4 EARLY/IMMEDIATE REMISSION
PROPORTION WITH EARLY/IMMEDIATE REMISSION The rate of recurrence after a second seizure is not influenced by antiepileptic drug therapy,(24,
175) although randomised trials have suggested that this may only be a long term outcome and
that medication may suppress seizures in the short term.(98, 108-111) The studies predicting
immediate remission employed different approaches to antiepileptic drug therapy: 2 studies(95,
96) had all the patients in the cohort placed initially on antiepileptic monotherapy; 1 study(98)
randomised treatment; the other 2 studies treatment did not randomise patient allocation to
medication (80% on medication in 1,(99) and 44% in the other(24)) with choice of therapy at the
discretion of the physician. These different approaches to treatment in the cohorts were not
reflected in the proportion of patients that achieved immediate remission. However, 3 factors
were important in influencing the proportion of patients that entered immediate remission:
case ascertainment, inclusion criteria, and the age of patients within the cohort.
Where there was exclusive prospective case ascertainment i.e. all the patients in the study were
identified at the exact point of their second seizure, fewer patients (37%) did not experience a
next seizure after 2 years(24) compared to between 42% and 57.5% in 3 of the 4 studies with
mixed case ascertainment,(95, 98, 99)especially including different proportions of first seizure,
prospectively identified and retrospectively ascertained patients. The proportion of the cohort
without immediate remission was highest (57.5%) where more than half of the patients were
single seizure patients.(98) In the other 2 cohorts, single seizure cases were less than a third of
the cohort. This supports previous evidence that for most patients, the rate of seizure
recurrence reduces with increasing number of seizures.(119, 301, 302)
However, among the cohorts with mixed retrospective and prospective case identification, 2
hospital based studies with similar proportions of patients in each seizure category (the ratio of
1st seizure to 2-5 seizures and >5 seizures being roughly 4:9:7 in both studies) had widely
33% in immediate remission at 2 years | 100% on AED | Del Felice et al 2010 37% in immediate remission at 2 years | 44% on AED | Shinnar et al 2000 42% in immediate remission at 2 years | 80% on AED | Lindsten et al 2001 48% in immediate remission at 2 years | 100% on AED |CGSEPI 1988 58% in immediate remission at 2 years | 50% on AED | Kim et al 2006
Box 16 | Proportion of patients in immediate remission at 2 years by the proportion of patients on AED
108
different proportion of patients in immediate remission at about 2 years of follow up: it was as
high as 48% in 1 (95) and only 33% in the other.(96) The hypothesised explanations for this
difference were:
1) The age difference between the cohorts. The cohort with 48% in immediate remission at 2
years after intake had a mean age of 19 years while the cohort with 33% had a mean age of
about 32 years suggesting that younger/childhood-onset epilepsy may have a better prognosis
than adult onset epilepsy,(48) which may be due to more symptomatic epilepsy occurring in
adult-onset epilepsy(303, 304) and benign syndromes occurring more in childhood(305).
2) Inclusion criteria in the cohorts. The study with 48% of the cohort in immediate remission at 2
years excluded patients with progressive neurological disorders associated with increased risk of
seizures e.g. brain tumour, while the other cohort included every patient with newly diagnosed
epilepsy.
The 2 studies (95, 98) that excluded patients with known poor prognosis compared to the other
3, had a higher proportion of patients in immediate remission at 2 years after intake (57.5% and
48%) at 2 years follow up. The study from a randomised controlled trial(98) (with 57.5%)
excluded all the patients for whom indication for antiepileptic drug therapy was certain, while
the other study(95) only excluded the patients with progressive neurological disorder. This
finding also suggests that patients with known poor prognosis, or “for whom indication to
commence treatment was not clear” are more likely to have a recurrence after diagnosis.
PREDICTORS AND NON PREDICTORS There were 2 consistent independent predictors of early/immediate remission in people with
newly diagnosed epilepsy: having 2 or more seizures before intake and a patient having remote
symptomatic epilepsy.
The patients with adult-onset epilepsy with 2 or more pre-intake seizures are 37% (95%CI -11%
to 64%) more likely than people with a single seizure before diagnosis to have a recurrence,
again showing that the risk of having a subsequent seizure may depend on the number of
previous seizures. However, the 95% confidence interval crosses unity for this estimate and is
therefore not particularly reliable. The second study that retained “having 2 or more pre-intake
seizures” was from an all age cohort with mostly young (mean age 19 years) patients. The study
however did not report the risk estimates. This predictor would anyway not have been available
in an ideal prognosis study of epilepsy, especially one that is predicting early remission, a
category in which the number of seizures being only 2 at intake into the cohort is particularly
important to limit potential for bias.
The children with remote symptomatic aetiology are 41% (95%CI 14% to 59%) more likely than
children with idiopathic/cryptogenic aetiology to have a recurrence.(24) The estimates from 2
109
other studies,(98, 99) while similar, may be fraught with the bias that the inclusion criteria of
the studies might introduce. The estimate from Kim et al,(98) an all age cohort which excluded
patients “for whom indication to commence treatment was not clear” and which also included
single seizure cases and people with more than 2 seizures prior to intake was: 26% (95%CI 6% to
42%). Lindsten et al,(99)also a mixed cohort, but of patients with adult-onset epilepsy, found
that adults with remote symptomatic aetiology are 66% (95%CI 23% to 74%) more likely than
people with idiopathic/cryptogenic aetiology to have a recurrence. These estimates suggest that
while remote symptomatic aetiology is a predictor of seizure recurrence in patients with newly
diagnosed epilepsy, it may be a more important predictor in adult-onset epilepsy.
The other variables – gender, age at onset of seizures, duration of epilepsy before intake into
cohort, abnormal EEG, seizure focus, family history and history of febrile seizures – assessed in
more than 1 study, were not found to be consistent independent predictors of early/immediate
remission in newly diagnosed epilepsy. That the age at onset of seizures is not associated with
recurrence suggests that the age difference in the proportion of people that achieve immediate
remission is accounted for by the fact that a higher proportion of patients with adult-onset
epilepsy have symptomatic aetiology (303, 304) and that benign epilepsy syndromes are more
common in childhood-onset epilepsy.(305)
5.5 REMISSION OFF ANTIEPILEPTIC MEDICATION
PROPORTION WITH REMISSION OFF ANTIEPILEPTIC MEDICATION Three study factors were important in explaining the variation in the proportion of patients that
were successfully weaned off antiepileptic medication: the study setting, the method of case
ascertainment, and the readiness of the treating physician to wean patients off medication. If
the proportions of patients in remission off medication at last follow up in the respective studies
are arranged according to the hypothesis that prospective, population based studies are the
least likely to be biased, and retrospective, hospital based studies are the most likely to be
biased,(20) the progression of the proportion of patients off AED is largely linear. (Box 17) The
less biased studies have a higher proportion of patients in remission off AED, except for 2 data
points from the Turku cohort (32) where the definition of remission off AED was such that the
patient had to have continued in remission off medication for at least 5 years. The relatively
lower proportion of patients in remission off AED may have been further accentuated by the
fact that 3 seizures instead of 2 diagnosed epilepsy in the Turku cohort and that the physicians
responsible for patients in the cohort were more reluctant to discontinue AED(105). This finding
however suggests that in childhood-onset epilepsy, more than two thirds of patients may be
successfully taken off antiepileptic medication.
110
*(Physicians were reluctant to discontinue AED)
The progression was not significantly affected by the fact that patients with generalised absence
seizures and known poor prognosis, including progressive neurological disorders were excluded
from the cohort with 54% in remission off medication(71) compared to 55% when the previously
excluded patients were included.(106) It is possible that the poor prognosis and the good
prognosis of generalised absence seizures balance each other in such a way that it did not
reflect in the proportions.
PREDICTORS AND NON PREDICTORS Two factors consistently predicted achieving remission off antiepileptic medication: intellectual
disability and having more than 1 seizure in the period between 6 and 12 months on
antiepileptic medication. Two types of model retained intellectual disability as independent
predictor of seizure-free status off medication: models that included only variables assessed at
intake/onset, and models that combined variables assessed at intake/onset with those only
assessable after 1 year of follow up such as having more than 1 seizure in the period between 6
and 12 months on antiepileptic medication.
The risk estimates from models with only intake/onset variables vary widely, but with
overlapping 95% confidence intervals. The model from the combined Dutch and reconstituted
Nova Scotia cohorts estimated that children with intellectual disability are 23% (95%CI 6% to
39%) less likely to achieve remission off medication than intellectually normal children.(106) The
estimate was much higher in the other 2 studies which were more prone to bias: 75% (95%CI
not stated) in Camfield et al(71) with generalised absence seizures and known poor prognosis
excluded, and 77% (95%CI 26% to 93%) in Oskoui et al,(72) a retrospectively ascertained,
hospital based cohort.
The risk estimates from models with both intake/onset and 1 year variables are also from these
2 more biased cohorts, showing that children with intellectual disability are 81% (95%CI 41% to
94%) less likely than intellectually normal children to achieve remission off medication in Oskoui
et al(72) and 71% (95%CI not stated) in Camfield et al. (71) However, the fact that the 4
estimates from 2 possibly biased studies are similar may be due to the fact that the true
52% Retrospective | Hospital | off AED at last follow up | Oskoui et al 2005 47% Mixed | Population |5 years off AED at last follow up | Sillanpaa et al 1998* 54% Mixed | Population |off AED at last follow up | Camfield et al 1993 55% Mixed | Population |off AED at last follow up| Geelhoed et al 2005 59% Mixed/Prospective | Population/Hospital |off AED at last follow up | Geelhoed et al 2005 65% Prospective | Hospital |off AED at last follow up | Geelhoed et al 2005 56% Prospective | Population |5 years off AED at last follow up | Sillanpaa et al 1998*
Box 17 | Proportion of patients in remission off AED arranged according to potential for bias
111
estimate of the risk of not achieving remission off medication may be closer to the estimate
from the study(106) that is less likely to be biased.
These same 2 potentially biased studies had models with both intake/onset and 1 year variables
which show that children with more than 1 seizure in the period between 6 and 12 months
while on antiepileptic medication are: 75% (95%CI not stated) and 76% (95%CI 40% to 90%) less
likely to achieve remission off medication compared with children who have 1 or no seizure 6 to
12 months on medication.
None of the other variables such as gender, age at onset, number of seizures before intake into
cohort, abnormal EEG, epilepsy syndrome and abnormal sign on neurological examination that
were assessed in more than 1 study, were consistently retained as an independent predictor of
remission off AED in children with newly diagnosed epilepsy.
EXTERNALLY VALIDATED MODELS Three models were externally validated and of the 3 models, 1 was derived with Cox regression
and 2 were derived by logistic regression. Neither of the logistic regression-derived models
reported the area under the receiver operating characteristics (ROC) curve, AUC, an important
measure of discrimination. However, the 2 logistic regression models reported the best
probability cut off (≈50%) at which the accuracy of the model was assessed and reported. None
of the 3 models was assessed for calibration.
The model that was derived using the reconstituted Nova Scotia cohort performed slightly
worse in the external validation study in the Dutch cohort.(106) Specificity reduced from 69% to
58% in external validation, while sensitivity increased marginally from 69% to 71%. This shows
that the model’s ability to detect negative outcome (no remission off AED) was poorer
compared to its ability to detect patients with remission off AED. The positive and negative
predictive values show that positive prediction was right 73% of the time, whereas negative
prediction was right in only 35% of the instances when it was made. However, the PPV and NPV
were expectedly higher in the validation study (PPV rises from 73% to 76%, NPV from 35% to
53%) owing to the fact that the pre-test probability (i.e. proportion of cohort with outcome,
being in remission off AED) was higher in the validation sample at 65% compared to 55% in the
derivation cohort.(279) This means that in the validation cohort, just knowing that a child has
epilepsy allows the prediction of remission off AED to be right 65% of the time. However, if the
prediction were to be based on the model and not prevalence, the positive predictive gain only
improves by 11%, compared to the negative predictive gain that rises by 18%. However, in spite
of these predictive gains, the model predicts wrongly in about 1 out of 3 children, although
worse in the external validation cohort (36% vs. 31%).
The second externally validated model that was derived using logistic regression was developed
on the Dutch cohort and validated on the reconstituted Nova Scotia cohort.(106) Specificity was
112
remarkably worse (reduced from 70% to 57%) in external validation with a marginal increase in
specificity (66% to 67%), showing that the model’s ability to detect negative outcome (no
remission off AED) was better compared to its ability to detect patients with remission off AED
in external validation. In spite of lower pre-test probability in the validation cohort (55% vs.
65%), the NPV slightly increases from 55% in derivation cohort to 56%. The PPV however
reduces from 79% to 67%. Therefore, in external validation, the positive predictive gain was 12%
while the negative predictive gain was 11% above pre-test probability. However, in spite of
these predictive gains, the model predicts wrongly in about 1 out of 3 children, although worse
in the external validation cohort (39% vs. 31%).
The model derived from the Nova Scotia cohort by Camfield et al(71) using Cox regression had a
marked loss in sensitivity when externally validated in the Turku cohort (74% to 47%), although
the specificity increased from 64% to 88%. However, owing largely to the pre-test probability
being higher in the validation cohort, (60% vs. 54%) the PPV increases from 71% to 84%,
although the NPV decreases to 51% from 67%. Therefore from a pre-test probability of 60% in
the validation cohort, the model increases the accuracy of positive prediction by 24% and of
negative prediction by 11%. The increase in positive prediction over pre-test probability may
however be due to the fact that instead of predicting remission off AED in external validation,
the model was used to predict remission on or off AED, a decision that was justified by the
argument that the physicians handling patients within the Turku cohort were reluctant to
discontinue medication and as 75% of those with 3 year remission at last follow up had been in
remission for at least 10 years, it was considered unlikely that these patients would relapse off
medication. However, the model predicts wrongly in about 1 out of 3 children, and worse in the
external validation cohort (39% vs. 32%).
The 3 externally validated models have not been evaluated prospectively in a randomised
clinical trial to assess their effect in clinical practice, possibly owing to their poor performance
and the fact that they only reported probabilities, which did not necessarily recommend any
particular course of action that may be assessed in a trial. Only the model from the Nova Scotia
cohort presented scores and the scores may be easy to use for physicians, as well as patients.
The other 2 models did not present scores for their models, and with the variables included,
(mostly of syndromic diagnosis) may not be quite as easy to assess for physicians and patients.
Finally, 2 of the 3 models (Nova Scotia intake and Nova Scotia reconstituted) had the most
consistent predictor variable, intellectual disability, in their model.
5.6 REMISSION ON OR OFF MEDICATION
PROPORTION WITH REMISSION ON OR OFF ANTIEPILEPTIC MEDICATION Four study factors were important in explaining the variation in the proportion of patients that
was reported to have achieved remission on or off medication among the cohorts and studies:
113
the definition of remission, the length of follow up, the study design, and the age of patients
within the cohort.
In studies that described their outcome as terminal remission (number of seizure-free months or
years immediately before the last follow up) the proportion of the cohort in remission varied
according to the length of seizure-free period that defined terminal remission: the proportion in
remission rises with length of follow up and falls with the length of terminal remission. When it
was 3 month terminal remission with 1 year minimum follow up, the proportion in remission is
as low as 50%.(93) It rises to between 73% and 76%, with 1 to 3 years terminal remission when
the minimum follow up ranged from 5 years to 20 years,(66, 68-70) and drops to 64% when 5
years seizure-free period defined terminal remission, even though the minimum follow up was
20 years.(32) For 2 largely prospective studies that defined terminal remission as 1 year seizure-
free period,(66, 70) the proportion of patients in remission remained at 76% in spite of doubling
(5 and 10 years) of minimum length of follow up between the 2 studies. (Box 18) These findings
show that it does take time for people with newly diagnosed epilepsy to enter terminal
remission; however, after a time period, possibly 5 years, patients who will ultimately enter
remission would already have achieved the outcome.
However, the result from Hui et al(97) a study of an adult cohort, had a lower proportion of
patients in remission relative to all the other comparable studies conducted in childhood onset
epilepsy. The proportion in remission for prospective studies was also higher than in comparable
cohorts with mixed and retrospective case ascertainment. This is in keeping with the evidence
that retrospective studies introduces bias to estimates of proportion of patients in remission(20)
and that childhood onset epilepsy may have a better prognosis than adult-onset epilepsy. (48,
303-305)
Eight studies defined remission by the seizure-free period during follow up, which may or may
not be terminal. Three (101, 104)of the 8 studies did not present the proportion of patients in
remission. Of the 5 that presented the proportion with outcome, as expected, the percentage of
patients in remission decreased as the number of seizure-free years that defined remission
increased. In these studies, the proportion of patients in remission was higher for the same
number of years when compared with those that entered terminal remission for the same
50% - 3 month TR with 1 year minimum follow up | Banu et al (Retrospective) 69% - 6 month TR with 2 years minimum follow up | Arts et al 1999 (Prospective) 60% - 1 year TR with 1 year minimum follow up | Hui et al 2007 (Mixed) 73% - 1 year TR (length of follow up not stated) | Lossius 1999 (Retrospective) 76% - 1 year TR with 5 years minimum follow up | Arts et al 2004 (Prospective) 76% - 1 year TR with > 10 years minimum follow up | Sillanpaa et al 2009 (Prospective) 76% - 3 year TR with > 20 years minimum follow up | Sillanpaa 1990 (Mixed) 64% - 5 year TR with > 20 years minimum follow up | Sillanpaa et al 1998 (Mixed)
Box 18 |Remission as seizure-free period immediately before last assessment i.e. Terminal Remission (TR)
114
number of years and with comparable length of follow up. This is also an expected pattern as
relapses may occur at any time to truncate terminal remission, whereas what is needed to
satisfy the definition “seizure-free period during follow up” is 1 stretch of the designated
seizure-free year(s) at any point during follow up.
One study (24)had a different measure for refractoriness (not achieving remission) which was
defined as having multiple recurrences of up to 10 seizures within about 8 years. In their cohort,
patients were initially identified at the point of first seizure, and the diagnosis of epilepsy
prospectively was confirmed with the second seizure. Of the patients with epilepsy 71% did not
have 10 recurrences after a mean follow up of about 8 years. Therefore the predictors
considered in this study included pre-diagnosis variables such as having the second seizure
(which confirms the diagnosis of epilepsy) within 1 year of the first.
PREDICTORS AND NON PREDICTORS There were 5 consistent predictors of achieving remission on or off antiepileptic medication:
having mixed seizure types at onset, symptomatic aetiology, intellectual disability, number of
seizures before index and the number of seizures in the first 6 months of follow up.
The finding from an all age cohort was that patients with mixed seizure types at onset are 30%
(95%CI -4% to 63%) less likely than patients with a single seizure type to achieve remission on or
off medication, although the 95% confidence interval crosses zero for this estimate and so may
not be reliable.(64) However, in childhood onset epilepsy, the odds of achieving remission are
about 1 in 4 [0.23 (95%CI 0.10 to 0.48)] for patients with multiple seizure types at onset
compared to those with a single seizure type at onset.(93) However, this estimate may also not
be reliable as it was from the study (93) that defined remission as 3 month terminal remission,
rather than much longer (e.g. 6 months to up to 5 years) as in the other studies that defined
remission as terminal remission.
None of the 4 studies (21, 24, 77, 104) that retained remote symptomatic aetiology in their
model defined remission as terminal remission. Three defined remission as “seizure-free period
during follow up” and 1 defined remission as not having up to 10 recurrences in an average of 8
years. However, the estimates from the 4 studies were similar. Of the 3 studies that defined
remission more conventionally, (as “seizure-free period during follow up”), 2 were in childhood-
onset epilepsy cohorts and of the 2, Berg et al(21) was the study with the less likelihood for bias
as it is a population based prospective study. Berg et al(21) estimate that it was 37% (95%CI 16%
80% - 1 year R with 1 year minimum follow up | Abduljabbar et al 1998 (Mixed) 78% - 1 year R (during the last 10 of min. 20 years) follow up | Sillanpaa 1993 (Mixed) 74% - 2 year R with 2 years minimum follow up | Berg et al 2001b (Prospective) 68% - 2 year R by 5 years of follow up | CGSEPI 1992 (Mixed) 51% - 3 year R by 5 years of follow up | CGSEPI 1992 (Mixed)
Box 19 | Remission (R) as seizure-free period during follow up
115
to 53%) less likely for children with remote symptomatic aetiology to achieve remission
compared to children with idiopathic/cryptogenic aetiology. The other study(77) was from a
mixed population based cohort, and the estimate was higher and with wider 95% confidence
interval at 56% (95%CI 8% to 75%). The adult-onset epilepsy study(104) also estimated that it
was 56% less likely for people with remote symptomatic aetiology to achieve remission
compared to those with idiopathic and cryptogenic aetiology, but without reporting the 95%
confidence interval. These estimates, particularly the similarity between the estimate from the
childhood-onset epilepsy study with mixed case ascertainment and a retrospective adult-onset
study suggest that the bias of the childhood-onset study may be towards a poorer prognosis for
achieving remission.
Intellectual disability was retained in 2 models: The odds of achieving remission are 4 in 10 [0.40
(95%CI 0.18 to 0.90) in children with intellectual disability compared to those with normal
cognitive development.(93) However, for adults-onset epilepsy, the odds are 1 in 10 [0.10
(95%CI 0.05 to 0.25)].(97) These results suggest that intellectual disability may be a more
important predictor of not achieving remission in adult-onset epilepsy compared to patients
who are diagnosed in childhood.
Two studies from different cohorts retained the natural logarithm of the number of seizures
before index in their models. For Arts et al(65)the risk of achieving remission reduced by 34%
(95%CI 5% to 54%) for every unit increase in the natural logarithm of the number of seizures a
child has before the index seizure at presentation, and for MacDonald et al(101) by 19% (95%CI
1% to 34%) for every unit increase in the natural logarithm of the number (plus 1) of seizures a
child has before the index seizure at presentation. This finding further suggests that “seizures
beget seizures”(49) could be true in that increasing number of seizures a child experiences
before intake/medication reduces the probability of achieving remission on or off medication.
The same studies also retained the natural logarithm of the number of seizures from the index
seizure to 6 months after intake/medication in their models, showing that early response to
antiepileptic medication is a predictor of eventual remission: For Arts et al(65)the risk of
achieving remission reduced by 50% (95%CI 28% to 65%) for every unit increase in the natural
log of the number (plus 1) of seizures a child has from the index seizure to 6 months after
intake/medication, and for MacDonald et al(101) by 41% (95%CI 30% to 50%) for every unit
increase in the natural log of the number (plus 1) of seizures a child has from the index seizure
to 6 months after intake/medication.
The other variables not consistently predictive of remission on or off medication include age at
onset, gender abnormal EEG at intake and the range of seizure types considered as were factors
like family history of epilepsy, history of neonatal seizures and the specific aetiological factors
considered.
116
EXTERNALLY VALIDATED MODELS Two models were derived in Arts et al 1999 by logistic regression(65) (one with only intake
variables, and the other including both intake and 6 month variables) were temporally validated.
(107)
The model with only intake variables was calibrated, with a plot of 4 risk groups of unequal size.
The calibration plot suggests that the model was well calibrated with the percentage of children
correctly predicted not to achieve remission increasing with the predicted chance of not
achieving remission, although calibration was not assessed using a formal statistical test. The
model however had poor discriminative ability (AUC <0.70) on both internal and external
validation. For the model, the chosen best probability cut-off was 38%: i.e. probability value
greater than 38% indicates not achieving 6 month terminal remission.(107)
The model performed slightly worse on external validation with sensitivity reducing from 62% to
60% in external validation, and specificity from 69% to 61%. This shows that the model’s ability
to detect poor outcome (no 6 month terminal remission at 2 years) was similar to its ability to
detect positive outcome (6 month terminal remission at 2 years) in the external validation
population.(107) However, the NPV increases as expected from 20% to 25%, but the PPV
reduces marginally from 47% to 45% in spite of the fact that the pre-test probability (i.e.
proportion of cohort with outcome, not achieving 6 month terminal remission) was higher in the
validation sample at 34% compared to 31% in the derivation cohort. Therefore, in the validation
cohort, the knowledge that a child is within that cohort allows for a prediction of a child not
entering remission to be correct 34% of the time (pre-test probability). However, if the
prediction is based on the model and not prevalence, the positive prediction improves by 11%,
compared to a loss of 9% in negative prediction. The model also predicts wrongly in more than 1
out of 3 children in internal and external validation. The negative predictive loss may be due to
the fact that the cut-off model was particularly set to enhance sensitivity at the expense of
specificity, thereby allowing for better prediction of not achieving 6 month terminal remission at
2 years.
The model with intake and 6 month variables was also calibrated, with a plot of 4 risk groups of
unequal size. The calibration plot suggests that the model was well calibrated with the
percentage of children correctly predicted not to achieve remission increasing with the
predicted chance of not achieving remission, although calibration was not assessed using a
formal statistical test. The model however had a fairly good discriminative ability (AUC >0.70) on
both internal and external validation. The best probability cut-off for the model was 34%: i.e.
probability value greater than 34% indicates not achieving 6 month terminal remission.(107)
The model performed worse on external validation with sensitivity reducing from 73% to 67% in
external validation, and specificity from 73% to 60%. This shows that the model’s ability to
detect poor outcome (no 6 month terminal remission at 2 years) was better than its ability to
117
detect positive outcome (6 month terminal remission at 2 years) in the external validation
population.(107) However, the NPV increases as expected from 14% to 22%, but the PPV
reduces from 55% to 47% in spite of the fact that the pre-test probability (i.e. proportion of
cohort with outcome, not achieving 6 month terminal remission) was higher in the validation
sample at 34% compared to 31% in the derivation cohort. Therefore, in the validation cohort,
the knowledge that a child is within that cohort allows for a prediction of a child not entering
remission to be correct 34% of the time (pre-test probability). However, if the prediction is
based on the model and not prevalence, the positive prediction improves by 13%, compared to
a loss of 12% in negative prediction. In spite of its positive predictive gain over pre-test
probability, the model also predicts wrongly in more than 1 out of 3 children in internal and
external validation. The negative predictive loss, as in the first model may be due to the fact that
the cut-off model was particularly set to enhance sensitivity at the expense of specificity,
thereby allowing for better prediction of not achieving 6 month terminal remission at 2 years.
Each of the models contained 2 of the 5 factors found to be consistent predictors of remission:
model with intake only variables (natural logarithm of the number of seizures before intake, and
remote symptomatic aetiology) and model with intake and 6 month variables (natural logarithm
of the number of seizures in the first 6 months after intake, and remote symptomatic aetiology).
They both presented scores for the assessment of risk, although the scores were complex, the
predictors many and the models may not be easily used by physicians and patients. The models
however have not been assessed in a randomised study to confirm their usefulness in a clinical
setting.
5.7 INTRACTABILITY
PROPORTION WITH MEDICALLY INTRACTABLE SEIZURES Two factors explained the differences in the proportion of patients with medically intractable
seizures in individual studies: how intractability was defined and the study design/setting.
Medically intractable seizures is usually defined as having “at least 1” or “more than 1” seizure
per month for about 1 year (ranged between at least 6 months and at least 2 years) after more
than 2 antiepileptic drug trials singly or in combination, at their maximum tolerable dose.(78)
However, studies that defined intractability as having “1 or more” seizures per month (72, 75,
76) had more patients who met the criteria for intractability (13% to 14%) compared to the
studies(73, 74) that defined intractability as only “more than 1” seizure per month (9% to 10%).
The were 2 outliers, both from retrospective hospital based cohorts: the one that defined
intractability as having recurrent seizures in the 6 months immediately before last follow up had
7% while the other, even though it defined intractability as having “at least 1” or more seizures
per month, one third (32.5%) of the cohort met the criteria for intractability. (78)
118
PREDICTORS AND NON PREDICTORS Five variables were consistently retained in models as independent predictors of having
medically intractable seizures: age at onset (continuous variable), onset of seizures in infancy
(age < I year), remote symptomatic aetiology, idiopathic aetiology and intellectual disability.
Three studies had models that retained age at the onset of epilepsy (as a continuous variable) as
an independent predictor of medical intractability. One of the studies(78) was a remarkably
biased retrospective hospital based study with about a third of the cohort having intractable
seizures. The second study (75) has an estimate that was not significant as the confidence
interval that crosses unity, hence unreliable. The third study(79) is also likely to be biased in
terms of study participation as there was also an over-representation of patients with
intractable seizures (being about 4 times as many as patients who are in remission).
However, when onset of seizures at less than 1 year was considered, 3 studies retained it in
their model. Ramos-Lizana et al(74)estimated that people with onset of seizures at infancy were
2.6 (95%CI 1.0 to 6.9) times more likely to have medically intractable seizures compared to
children whose seizures begin after the age of 1. The estimate is remarkably similar at 2.6
(95%CI 1.1 to 4.30) in Casetta et al(75). The third estimate is however different from the other 2,
possibly because the study by Chawla et al(94) may be particularly prone to bias. In the quality
appraisal for potential for bias, it was unlikely to be biased in only 1 area of potential bias
(multivariate analysis) out of 5 (others being study participation, study attrition, prognostic
factor measurement and outcome measurement). For the study,(94) the estimate was that
people with onset of seizures at infancy were 3 (95%CI 2.0 to 3.6) times more likely to have
medically intractable seizures than children whose seizures begin after the age of 1.
Of the 4 models that identified remote symptomatic aetiology as an independent predictor of
medically intractable seizures, 3 were particularly prone to bias. Two (78, 79) of the 3 were
estimates from cohorts within which patients with intractable seizures were significantly over-
represented. The third study was Chawla et al(94) which was unlikely to be biased in only 1 out
of 5 areas of potential bias assessed in the studies included in this review. The 3 estimates from
bias prone studies were similar with relative risk between 1.34 and 2.06 and largely overlapping
confidence intervals. However, Casetta et al(75) estimates that it is about 5.48 (95%CI 2.10-
07% - “6 months recurrent” | Retrospective, Hospital | Oskoui et al 2005 09% - “At least 1 Seizure” | Mixed, Hospital | Ramos-Lizana et al 10% - “At least 1 Seizure” | Prospective, Population | Berg et al 2001a 13% - “More than 1 Seizure” | Retrospective, Hospital | Oskoui et al 2005 14% - “More than 1 Seizure” | Prospective, Population | Casetta et al 1999 14% - “More than 1 Seizure” | Mixed, Population | Kwong et al 2003 33% - “At least 1 Seizure” | Retrospective, Hospital | Berg et al 1996
Box 20 | Influence of definition and study design/setting on the proportion of patients with intractable seizures
119
9.64), indicating that children with remote symptomatic aetiology are 5.5 times more likely to
have medically intractable seizures than those with idiopathic/cryptogenic aetiology.
Three studies identified idiopathic aetiology as an independent predictor of intractable seizures
in that children with idiopathic aetiology are less likely than those with symptomatic aetiology
and cryptogenic aetiology to develop intractable seizures. However, 1 of the studies had an
atypical definition of intractability (recurrent seizures in the 6 months before last follow up or
requiring surgery or ketogenic diet to control seizures); the study estimates that it is 87% (95%CI
48% to 97%) less likely for a patient with idiopathic epilepsy to develop intractable seizures
compared to children with symptomatic/cryptogenic aetiology. Estimates from the other 2
studies estimate are similar: 80% (95%CI 20% to 100%) less likely and 77% (95%CI 37% to 91%)
less likely to develop intractable seizures with idiopathic aetiology than children with
symptomatic/cryptogenic aetiology.
Intellectual disability was also shown to consistently predict medically intractable seizures. The
odds of having intractable seizures are 7 times [7.2 (95%CI 1.0-50.8)] in 1 study,(72) albeit with
EPV less than 10, and much higher in the other study which is less likely to be biased(76) at 18
times [18.2 (95%CI 5.2-63.6)] greater for children who have intellectual disability and/or
cerebral palsy relative to those with normal cognitive development and without cerebral palsy.
However, the confidence interval is remarkably wide for both estimates, which is probably due
to few children in both studies having intellectual disability.
However, other potential predictors such as mixed seizure types, specific seizure types, seizure
frequency, abnormal EEG, occurrence of status epilepticus, and family history of epilepsy or
history of neonatal seizures, and were not found to be a predictor of intractability.
5.8 REMISSION AFTER RELAPSE There was only 1 study(25) in this category. This fact precludes much further discussion.
However, the study showed that the odds of not achieving remission after a relapse are about 8
times as high in children with remote symptomatic aetiology compared to those without
symptomatic aetiology.
5.9 IMPLICATIONS FOR CLINICAL PRACTICE The results of this review suggest that in deciding whether to initiate antiepileptic drug therapy
in patients with newly diagnosed epilepsy, particular consideration should be given to whether
the newly diagnosed patient is a child or an adult, if the epilepsy has asymptomatic aetiology
and also possibly the occurrence of more than 1 seizure before the index seizure.
In making the decision, or in advising patients already in remission on whether to discontinue
antiepileptic drug therapy, it may be important for physicians to be more reluctant or more
120
careful in patients with intellectual disability and those who had more than 1 seizure during the
period between 6 and 12 months while on medication.
Patients with mixed seizure types at onset, remote symptomatic aetiology, intellectual disability,
high number of seizures before diagnosis and poor early response to medication indicated by
more than 1 seizure in the first 6 months of antiepileptic medication may require a more
aggressive treatment strategy to prevent their seizures from becoming refractory to
antiepileptic medication.
The children with onset of seizures in infancy (age < I year), with remote symptomatic aetiology,
and intellectual disability, unlike children with idiopathic aetiology may require a more
aggressive treatment strategy to prevent their seizures from becoming medically intractable and
to also be managed with a view to early consideration of epilepsy surgery within 2 years of
diagnosis. The parents and relatives of the children with these risk factors, especially when they
occur together in the same child may need to be informed early on in the course of the illness
regarding the possibility of intractability and the management strategies that may be necessary
in addition to pharmacological intervention in case of intractability.
In all, there are at present no satisfactory prediction models for any of the outcome categories.
It may suffice however to inform and advise patients and their relatives based on the proportion
of patients that achieve each outcome. These predictors may be all that is presently available for
the purpose of devising management strategy early in the course of the disease and for advising
patients.
5.10 DIRECTIONS FOR FUTURE RESEARCH This segment presents the suggested directions and recommendations for future studies related
to this systematic review. These issues are discussed under the sub-headings based on the key
areas of this review: the literature search, reporting characteristics and quality appraisal of
studies, the classification of studies included in this review and the results of the review of
prognostic factor studies in newly diagnosed epilepsy.
THE LITERATURE SEARCH Four recommendations are made based on findings from the literature search: 1) There is need
for more evidence on the number and exact databases that would be necessary to search in
order to identify observational studies. Previous research on this issue has focused on
identifying randomised trials. 2) Future systematic reviews of studies containing multivariate
analysis may benefit from concentrating their search period to say the last 20 years. 3) Future
systematic reviews of observational studies in epilepsy may benefit from hand-searching recent
editions of Epilepsia especially where the resources are available, and the search does not
include EMBASE as the 4 publications unique to EMBASE were actually published in 2009 and
121
2010 in MEDLINE-indexed journals. 4) The clinical uptake of the results of future studies of
prognosis in newly diagnosed patients with epilepsy may benefit from authors and journal
editors choosing to publish some of those studies in non-specialist and primary care journals.
QUALITY AND REPORTING CHARACTERISTICS Four recommendations are made based on findings from quality appraisal and reporting
characteristics of included studies: 1) To enhance the robustness of multivariate analyses, it may
be necessary to engage in international collaboration studies in order to boost the number of
patients included in analyses. This may have the added advantage of rare syndromes being
better represented in multivariate analyses to better ascertain their prognostic significance, and
for including studies based in Africa and Australasia as none of the studies in the review
included patients from the 2 continents. 2) It would be beneficial in assessing the quality of
future observational studies and for future systematic reviews if authors and journals adhered
to the STROBE checklist as minimum standard for reporting observational studies. 3) It may also
be possible therefore for a future systematic review to compare the reporting characteristics of
publications in STROBE-endorsing journals with journals that leave the reporting of
observational studies to the discretion of authors. 4) The review, especially of studies predicting
medical intractability, shows that studies with a higher potential for bias according to the quality
items used in this review have risk estimates that differ in most cases remarkably from studies
with a lower potential for bias. There is need for further research into ways of adapting the
Hayden et al criteria to reviews of prognosis studies in other subject areas.
THE CLASSIFICATION OF STUDIES The classification scheme developed in this thesis has a potential advantage for future studies
because study categories were presented with the prognostic sub-groups (i.e. potential seizure
outcome categories) used to define and delineate comparison groups. It would be a task of
future studies of seizure outcome of newly diagnosed epilepsy to determine the proportion of
patients in each sub-group. To ease and facilitate subsequent systematic reviews and meta-
analyses, authors of future studies of seizure outcome in patients with newly diagnosed epilepsy
may locate their study within this classification scheme or its extension and label them as such.
This will also allow results of the studies to be interpreted more easily and readily within the
context of previous studies in the same category.
PROGNOSTIC FACTOR STUDIES IN NEWLY DIAGNOSED EPILEPSY
Further studies of immediate remission in newly diagnosed epilepsy will benefit from ensuring
prospective case ascertainment at the point of a second seizure as this is the most important
factor determining the proportion with outcome in this category of studies of seizure outcome
in newly diagnosed patients with epilepsy.
The deliberate inclusion of the consistent predictors of outcome across seizure outcome
categories, especially remote symptomatic aetiology and intellectual disability and of 1 year
122
variables, especially of the occurrence of seizures within the first year after diagnosis or while on
medication may also be important in studies predicting remission off medication, and remission
on or off medication.
The fact that intractability is a rare outcome, occurring in less than 15% of cohorts makes it
particularly important for future studies to ensure that sample size is adequate to build the
model with at least 10 events per independent variable entered into the model.
There should also be studies investigating predictors of seizure outcome in patients with adult-
onset epilepsy as most the studies eligible for this review were in childhood-onset epilepsy.
The distinguishing characteristic of the only study that investigated remission after relapse is
that it is a cohort with long follow up, ranging from 11 to 42 years.(25) It would require such
long follow up to determine patients who will relapse, and then following relapse achieve
terminal remission or not achieve terminal remission. Therefore, it is hoped that presently
existing cohorts will continue to be followed so that we can better understand the
characteristics that influence achieving remission after relapse(s) and of other seizure outcome
categories.
5.11 STRENGTHS AND WEAKNESSES OF THE REVIEW This review of prognosis studies in unselected populations of patients with newly diagnosed
epilepsy has accomplished its aim of classifying studies, exploring heterogeneity, and identifying
consistent predictors of seizure outcome in patients with newly diagnosed epilepsy. The main
strengths of this thesis are its thorough examination of studies included for their potential for
bias and exploration of sources of heterogeneity. The review was focussed specifically on
studies with multivariate regression analysis because they are best suited to control for possible
confounding bias(23). The thesis has also explored other potential sources of bias, using
objectively identified and clearly defined criteria.
The studies included in the review were also classified such that like was compared to like within
the study categories. The study also fills an important gap in the literature as there has been no
previous review of the methods and results of prognosis studies that identified independent
predictors of seizure outcome in newly diagnosed epilepsy. The quality appraisal items for
systematic reviews of prognosis studies developed by Hayden et al(39) and adapted for use in
this review have not been tested for validity and reliability. This review provides an instance
where the quality items have been adapted for use in a specific subject area.
However, the work has several limitations and future discussion on the methods of identifying
and handling consistent independent predictors in systematic reviews and meta-analysis is
needed. There is a potential drawback to predictors identified by multivariate analysis,
especially when it is in studies not aimed at investigating the predictive value of a particular
123
prognostic variable, and when a stepwise algorithm is used in selecting variables to include in
the model.(45) It has been demonstrated through multiple bootstrap and split-sample analyses
that the models so derived have low reproducibility within the same study sample.(306)
However, the strategy this review employed to mitigate potentially spuriously identified
predictors was to stipulate from the outset that only predictors identified in more than 1 study
conducted by different groups on different cohorts will be considered to be consistent
predictors of a particular outcome.
The discussions particularly relating to exploring factors that may explain the variations in the
proportion of patients with seizure outcome in each category of studies is limited by the fact
that only studies with multivariate analysis were included in the review. There are more studies
in the literature that may have determined these proportions and might have thus provided
more material for the discussions. However, the sample of studies in each category provides a
heterogeneous sample of studies. Therefore there were enough studies on which to base the
exploration and discussions in order to understand the characteristics that explain biases in the
studies in each category.
The Zhang-Yu equation(266) was used where possible to convert the odds ratios from logistic
regression analyses in order to approximate relative risk. The equation allowed for comparison
of odds ratio with the risk estimates (hazard ratios) from Cox regression analyses, which was
assumed to be a good approximation of relative risk. However, the conversion of odds ratio to
relative risk is in itself only an approximation. (307, 308) The formula has been shown to
account for only 85% of the required adjustment of odds ratio towards the relative risk. The 95%
confidence intervals calculated using the formula are also much narrower, with the results from
the formula being only about two thirds (67%) of the appropriately determined 95% confidence
interval of the relative risk.(308) However, the Zhang-Yu equation provides a useful approach to
interpreting risk estimates from logistic regression in a way that enhances comparison with the
relative risk which is more intuitive to understand. There is a need to use alternatives to the
odds ratio because of the limitations(308) of the Zhang-Yu equation.(266) Results of simulation
studies have shown that there are viable alternative models to the logistic regression
model.(309, 310) These models (Poisson regression and log-binomial regression) better
approximate the relative risk and could be used on longitudinal data with binary outcomes.(309)
There is a possibility of selection bias in this systematic review. The prognosis studies that did
conduct multivariate analysis to identify independent predictors may be of higher quality. There
may be publication bias as studies with significant results (i.e. that identified independent
predictors) may be more likely to get published. The search strategy yielded reviews published
in English-language, peer-reviewed, MEDLINE- and EMBASE-indexed journals with adequate
keywords and Medical Subject Heading (MeSH) or Emtree terms. There was no search of the
grey literature. The methods used to identify publications may have missed some eligible, likely
lower quality publications. However, given time and budget constraints, it was not feasible to
search the grey literature or to translate non–English-language publications. Indeed, given that
124
the search is likely exhaustive within its limited framework, these factors are unlikely to have
had much impact on the results and recommendations of this review.
For the same purpose of time and budget constraints, the data extraction, quality appraisal and
calculations were conducted by 1 reviewer. This is a potential source of error and bias in the
systematic review. However, the reviewer (MPhil candidate) extracted the data directly from
the relevant publications 3 times while checking against the previous extraction at each stage to
ensure accuracy. The quality appraisal was also conducted transparently and the information
upon which quality appraisal was based was explicitly reported in the review such that
researchers and practitioners can independently assess the quality of the studies included in the
review. There was also limitation due to incomplete reporting by authors as only published data
was used and the MPhil candidate did not contact authors to obtain additional information due
to time constraints. However, the candidate searched referenced papers and also publications
from the same group or cohorts for additional information where such information may be
important for the quality appraisal.
Indeed, this work has several limitations. Future research should continue to discuss, debate
and explore bias in prognosis studies and how to identify independent predictors of outcomes
from studies conducted by multivariate analysis. This will further develop, expand and establish
this burgeoning area of interest.
125
6 CHAPTER SIX: CONCLUSION This thesis has shown that although a wide range of variables have been considered across the
different seizure outcome categories, only a few have been consistently confirmed as being
statistically significant independent predictors in multivariate analysis. The study demonstrates
the feasibility of systematic review with thorough quality appraisal as a means of identifying the
consistent predictors of an outcome in studies that do not specifically investigate one particular
prognostic variable. Table 24 summarises the independent predictors and presents the least
biased estimate for childhood-onset and adult-onset epilepsy from each study category:
1.) Having more than 1 seizure before intake and remote symptomatic aetiology were
positive predictors of recurrence of seizure after intake (i.e. no immediate/early
remission) in childhood-onset and adult-onset epilepsy.
2.) Having more than 1 seizure in the period between 6 and 12 months on medication and
intellectual disability were negative predictors of achieving remission off medication in
childhood-onset epilepsy. None of the studies in this category considered adult-onset
epilepsy.
3.) Having more than 1 seizure before intake, having seizures in the first 6 months after the
index seizure, mixed seizure types at onset of epilepsy, intellectual disability and remote
symptomatic aetiology were negative predictors of achieving remission on or off
medication in both childhood-onset and adult-onset epilepsy.
4.) Having onset of seizures in infancy, intellectual disability, and remote symptomatic
aetiology were positive predictors of medical intractability, while idiopathic aetiology
was a negative predictor of intractability in childhood-onset epilepsy. None of the
studies in this category considered adult-onset epilepsy.
The neurobiology of epilepsy is heterogeneous. Therefore the aetiological classification, number
of seizures, mixed seizure types, and comorbidity with intellectual disability that feature among
independent predictors of seizure outcome may not fully capture the details of the biology, and
possibly the prognosis of epilepsy. Important as the independent predictors are, they may not
be important in the consideration of treatment and interventions in individual patients. For
example, some types of seizure manifestation (e.g. focal seizures without impaired
consciousness, absence seizures or seizures occurring only during sleep), and the benefits of
treatment may not outweigh the potential adverse effects of medication. When seizures occur
infrequently, even when the seizures are of a more severe form, the benefits of treatment over
adverse effects also have to be weighed on a case by case basis.(8) Therefore these predictors
can only serve as flexible guides to treatment and overall management strategy.
126
Table 24 | Consistent early predictors of seizure outcome in newly diagnosed epilepsy
IMMEDIATE REMISSION
REMISSION OFF MEDICATION REMISSION ON OR OFF MEDICATION (REMISSION)
MEDICAL INTRACTABILITY
ONSET OF SEIZURES IN INFANCY (AGE < 1YEAR)
RR 5.48 (95%CI 2.10-9.64) C
MORE THAN 1 SEIZURE BEFORE INTAKE
RR NOT STATED CA RR 0.63 (95%CI 0.36-1.11) A
RR 0.66 (95%CI 0.46-0.95)* C RR 0.81 (95%CI 0.66-0.99)* CA
NUMBER OF SEIZURES FROM INTAKE TO 6 MONTHS
RR 0.50 (95%CI 0.35-0.72)‡ C RR 0.59 (95%CI 0.50-0.70)‡ CA
NUMBER OF SEIZURES, 6 TO 12 MONTHS ON MEDICATION
RR 0.24 (95%CI 0.10-0.60) C
MIXED SEIZURE TYPES AT ONSET
RR 0.70 (95%CI 0.37-1.04) CA OR 0.23 (95% CI 0.10-0.48) A
INTELLECTUAL DISABILITY RR 0.77 (95%CI 0.61-0.94) C OR 0.40 (95%CI 0.18-0.90) C OR 0.10 (95%CI 0.05-0.25) A
OR 18.2 (95%CI 5.2-63.6) C
REMOTE SYMPTOMATIC AETIOLOGY
RR 0.59 (95%CI 0.41-0.86) C RR 0.44 (95% CI 0.26-0.77) A
RR 0.63 (95% CI 0.47-0.84) C RR 0.44 (95% CI NOT STATED) A
RR 5.48 (95%CI 2.10-9.64) C
IDIOPATHIC AETIOLOGY
RR 0.20 (95%CI 0.0-0.80) C
RR-Relative Risk, OR-Odds Ratio, C-Estimate from childhood-onset epilepsy, A-Estimate from adult-onset epilepsy, CA-Estimate from cohort with both childhood-onset and adult-onset epilepsy *Relative risk of the natural logarithm of every additional seizure before intake ‡Relative risk of the natural logarithm of every additional seizure in the first 6 months of follow up.
127
The findings of this systematic review, especially of lack of association of several factors with seizure
outcome, is in keeping with a dated systematic review by Berg and Shinnar in 1991(44) which identified
factors associated with recurrence after first seizure, although mostly in studies with univariate analysis.
They found, as in this thesis, that age at onset, gender, abnormal EEG, family history of epilepsy, history
of neonatal seizures, status epilepticus or febrile seizures, antiepileptic medication were not consistently
predictive of seizure outcome (recurrence after a first seizure). However, they also found that aetiology
(remote symptomatic increasing the risk for recurrence and idiopathic with lower risk), was a consistent
predictor of recurrence as was seizure type (focal seizures tend to recur after a first).
For Berg and Shinnar(44) as well, the method of case ascertainment was an important source of
discrepancy in the proportion of patients with outcome as retrospective and mixed case ascertainment
tend to result in biased estimates. In this thesis, the exploration of potential reasons for discrepancies in
the proportion of patients with seizure outcome in each category was an illuminating exercise which
yielded much information in assessing potential of individual studies for bias, and the differences in
outcome between studies of childhood-onset epilepsy and those of adult-onset epilepsy. The adoption
of this strategy in further systematic reviews of outcome in epilepsy is recommended.
Part of the sources of discrepancies in proportion of patients within cohorts with particular outcomes
was the specific details of how seizure outcome measure was defined in each study. For example,
studies that defined remission as terminal remission had fewer patients in remission relative to
comparable studies that defined remission as seizure-free period during follow up. Therefore it is
important to qualify estimates specifically by how they were defined in the study, the design of the
study, the setting of the study, and the patient population in the study when quoting proportion of
patients with any seizure outcome in epilepsy. It is also recommended that a future task for ILAE
(International League Against Epilepsy) in updating the guidelines for epidemiological research in
epilepsy(1, 2) will be to define to the specific details, the outcome measures (e.g. remission and
intractability) that studies within the categories identified in this thesis will have to adhere.
The current models for predicting seizure outcome in newly diagnosed patients with epilepsy are
neither accurate, nor have been rigorously developed and externally validated. The predictors identified
in this thesis may therefore be the only reliable data currently available to allow the easy identification
of those patients at the greatest risk of experiencing poor seizure outcome when newly diagnosed with
epilepsy. Further research in this area would be of considerable importance not only in terms of
increasing the understanding risk factors for poor outcome in epilepsy, but also in advancing health care
delivery to improve patient care and reduce the burden of seizures on quality of life.
As challenging as they are to conduct, population based studies with prospective case identification at
the point of second seizure remain the ideal study design for seizure outcome in epilepsy. More of these
studies will need to be conducted, especially among the 3 populations that were not well represented in
the cohorts on which the studies that made up this thesis were based: populations of people with adult-
onset epilepsy, people in Australasia and people in developing countries, especially of Africa, as about
80% of epilepsy patients live in developing countries(3).
128
129
7 APPENDICES
7.1 Appendix I: PROTOCOL EARLY PREDICTORS OF SEIZURE OUTCOME IN NEWLY DIAGNOSED EPILEPSY – SYSTEMATIC REVIEW OF PROGNOSIS STUDIES INTRODUCTION Prognosis studies investigate the relationship between occurrences of outcomes and predictors in defined populations of people with disease.(1, 2) The importance of prognostic research in epilepsy as in other chronic illnesses/diseases is overwhelming. They help to improve the understanding of the disease process, to guide treatment decision making (if and when to commence antiepileptic medication, if and when to have epilepsy surgery and other non-pharmacological interventions), and to predict the outcome of epilepsy more accurately for the purpose of patient information and counselling. (3) However, in spite of the methodological demands, prognostic studies are usually not protocol driven, small sized, widely heterogeneous in characteristics of patients, choice and number of predictor variables, outcome and follow up measures, and often prognostic analyses are improvised as icing on the cake of studies designed for quite another purpose. For these reasons, the results of prognostic research are largely fraught with limitations, with attendant uncertainty about the reliability of conclusions of their synthesis/meta-analyses. (1) This has led to the creation of a new Cochrane Prognosis Methods Group in 2008 aimed at providing support and a forum for discussion to facilitate and improve the quality of systematic reviews of prognosis research.(4) Systematic review of prognosis studies with particular focus on methodology is therefore a burgeoning field of interest, and there have been efforts at ensuring the quality of such reviews. Hayden et al have developed a set of criteria for quality assessment through a meta-review of systematic reviews of prognosis studies. (5) There have only been systematic reviews in the literature of epidemiological studies that focus on incidence and prevalence of epilepsy generally(6) and with regional focus - Asia(7), Latin America(8), Middle East and North Africa(9), Europe(10) and sub-Saharan Africa(11). However, these studies, when they do, only make passing reference to the methodology and/or results of prognosis studies. Ross et al also systematically reviewed the literature, but on management issues in epilepsy from 1980 to 1999.(12) The studies on prognosis were therefore only partly considered and only in passing, without methodological rigour. Parenthetically, also no systematic review has considered other outcomes in prognosis studies, important as they are: psychopathology outcomes (depression, anxiety, et cetera), neurodevelopmental outcomes (especially IQ in children), mortality, and quality of life measures. This will be the focus of future systematic review(s) following completion of the present one that is being proposed. However, an epidemiological synthesis of the natural history of epilepsy by Kwan and Sander(13) suggest that there are three prognostic groups: group 1 (20–30%) comprises patients with excellent prognosis, as there is long term remission after a period of seizure activity, with or without antiepileptic drug treatment and the primary aim of antiepileptic drug treatment in this group of patients is to suppress seizures until ‘spontaneous’ remission occurs, and patients can be successfully withdrawn after
130
a period of seizure freedom; for group 2 (20–30%) seizure remission occurs only with treatment and patients only remain seizure-free with continuing antiepileptic drug treatment; and in group 3 (30–40%) there is continuing seizures despite antiepileptic drug treatment with some patients having frequent debilitating seizures qualifying them as having ‘refractory’ epilepsy. Significance Seizure outcome in epilepsy is varied. Epilepsy itself is a multi-aetiological and diverse disorder. The first seizure brings with it the opening of a fresh chapter in the life of the patient, with tough decisions to be made for the patient and for clinicians especially of when or whether to start medication as they carry risks of acute idiosyncratic reactions, dose-related and chronic toxic effects, and teratogenicity. However, for most patients diagnosed with epilepsy, the benefits of treatment will far outweigh the risks associated with treatment, but for those who have had a single seizure and for those who have seizures with minor symptoms, this risk to benefit ratio is more finely balanced.(14) There is also the group of patients (Kwan & Sander’s Group 3)(13) that may be candidates for epilepsy surgery as they will develop debilitating drug-resistant epilepsy; as seizures themselves are not benign events, there is the consequent considerable clinical and psychosocial distress or even mortality. There is thus the need for studies that attempt to identify independent predictors of these seizure outcomes and also therefore useful to review the methodology and possibly synthesise the results of studies that have identified the factors which best predict clinically relevant seizure outcomes in patients that have been newly diagnosed with epilepsy. There is a greater tendency for publication bias in observational studies compared to randomised clinical trials (15) and prognosis studies particularly so, as it is probable that studies showing a strong, statistically significant prognostic ability are more likely to be published and this can lead to invalid results and incorrect inferences.(16) The proposed review will therefore thoroughly assess for quality based on indices that reflect how prone the studies are to bias. Objectives This thesis aims to answer the following questions: 1.) What is the quality of prognosis studies that have attempted to identify independent predictors of seizure outcome among unselected population patients with newly diagnosed epilepsy? 2.) What is the effect of study quality, especially potential risk for bias, on the results of the prognosis studies? 3.) Which factors have been consistently identified as independent predictors of seizure outcome and which factors have been consistently identified as non-predictors? 4.) Do satisfactory seizure outcome prediction models already exist? The review also proposes to assess the state of evidence of the prognosis of seizure outcomes in epilepsy with the view towards a possible synthesis of the results. A systematic review of methodology may also help in identifying areas of potential limitation and specific actions that may improve future investigations of prognosis in epilepsy. METHODS The guidelines for the design, performance and reporting for meta-analyses of observational studies published by the MOOSE group (17) will be followed in this systematic review.
131
Eligibility Criteria The review will seek for inclusion, published cohort, nested case control and case control studies of unselected population of people with epilepsy that assess the independent effect of 3 (an arbitrary decision, following the example of Counsell and Dennis(18), based on the fact that 2 variables will be too few to give significant information about their independent effect) or more predictor variables on a range of measures of seizure outcome in patients with epilepsy collected within the first year of onset, diagnosis, or presentation, with patients followed up for at least 1 year. Only studies published in English will be included in the review. There is no consensus on sample size estimations for multivariable models. However the standard rule of thumb is 10 or more events per variable entered into the model to allow a robust estimation of the coefficients (19), although a recent study showed that this number could be lower (20). Since about 30-40%(13) of patients with epilepsy do not achieve seizure remission despite continued AED treatment, a study of association of 3 predictor variables with this outcome will require at least 100 patients to achieve an EPV of 10 or more. Therefore we will also be seeking studies that include at least 100 patients. Studies with fewer than 100 patients will only be included if they have an EPV greater than 10 or have been validated on other data sets. Seizure outcome has been assessed using different concepts of measure. However, the review will not be limited to outcomes assessed in these forms only: Terminal Remission (length of seizure-free period immediately before last follow up evaluation); Longest Remission (longest seizure-free period during follow up); Time to Remission (time from diagnosis to the commencement of remission); Time to Recurrence (time from diagnosis to recurrence of seizures); Seizure Frequency (number of seizures per specified period of time); Refractoriness (lack of response to antiepileptic medication); and Intractability (lack of response to seizures with debilitating frequency of seizures). Search Strategy There is probably no widely acknowledged optimal strategy for searching the literature for prognostic studies.(3) Wilczynski et al (21)have developed search strategies that optimise the yield for prognostic studies in MEDLINE(21) with good sensitivity (the proportion of high quality articles that are retrieved for a particular topic) and specificity (the proportion of low quality articles not retrieved), although understandably low precision (proportion of retrieved articles that are of high quality) as searches were not limited by clinical content terms. The most sensitive search strategy is recommended for those interested in all articles reporting studies on prognosis and who are willing to sort out less relevant articles. The search strategy developed using Ovid's search engine syntax for combination of terms with the best sensitivity was: Incidence (MeSH) OR explode mortality OR follow-up studies (MeSH) OR prognos* (text word) OR predict* (text word) OR course (text word) These results will serve as a guide in deciding the search strategy for the review. These results were from searches not limited by specific clinical/disease terms and the authors suggest rather guardedly that it may be possible to increase the performance measures by combining search strategies with content specific terms using the Boolean 'AND'. Therefore MEDLINE was searched (from inception to March 2010) using “epilepsy” or “seizure” or “seizure disorders” with the explode feature where applicable, in combination with the search terms for prognostic studies developed by Wilczynski et al (21).
132
Furlan et al(22) identified terms in EMBASE related to study design and analysis that could reverse identify non-randomized studies in 4 systematic reviews across 4 different clinical areas. They found that text word “multivariate” was 1 of 2 terms which limit topic only searches in all 4 clinical areas. The text word “regression” (Cox regression and logistic regression) was among others common to 2 of the 4 clinical areas. These terms identified by Furlan et al(22) focusing on terms related to study design and statistical analysis will be used for the EMBASE search. The terms “multivariate” and “regression” are common to the inclusion criteria for studies to be included in the review. The terms were thus selected, and they will be used in addition to “multivariable” to limit an “epilepsy” topic search [explode epilepsy (Emtree) OR epilepsy (text word)]. The search will also be from inception to March 2010, and will not be limited to EMBASE-specific records alone. The results of the MEDLINE and EMBASE search will be screened after they have been exported to the EndNote citation manager. There will be a three-step selection process to identify eligible studies. In the first step, the title and abstracts will be screened to identify all studies that are neither about prognosis nor epilepsy and do not meet any of the inclusion criteria, duplication or redundant studies, papers that are review, editorial, commentary or letter, studies with non-seizure outcomes, studies with population restricted to a particular epilepsy subtype or patient population (e.g. surgical cohort, drug withdrawal cohort et cetera), studies with epilepsy or seizures as an outcome of or in relation to another pathological process and exclusively first seizure studies. These publications will be excluded, and will then be left with only the potentially eligible papers. In the second step, after reading the full text versions of the remaining studies, those without multivariate analysis of predictor variables for seizure outcomes in unselected cohort of people with epilepsy will again be excluded. Then in the third step, there will be a manual search the reference lists of all eligible publications. Data Extraction The data will first be extracted from each paper directly into a form, then from the paper also directly onto an Excel spread sheet database; results of the 2 rounds of data extraction will be compared for accuracy, and where there are discrepancies, the paper will be consulted for clarification. The extracted data will include authors and title of study, year of publication, study design, study size, age range and sex of the participants, predictor variables assessed, investigated seizure outcome measures, and the independent predictors with their risk estimates and corresponding 95% confidence intervals. All data will be extracted from the published studies and their authors will not be contacted for further information. REVIEW OF METHODS OF STUDIES In a meta-review of systematic reviews of prognosis studies, Hayden et al (5) developed extensive guidelines for assessing quality in prognosis studies on the basis of a framework of potential biases. Hayden and colleagues set out by identifying quality items in systematic reviews of prognosis studies, and subsequently pooling the items identified into 6 areas of potential bias (Study Participation, Study Attrition, Prognostic Factor Measurement, Confounding Measurement, Outcome Measurement and Analysis) all of which are relevant to the purpose and question of the present systematic review and will therefore be adapted in the analysis of methodology of the studies being reviewed. (Appendix IV) Studies with Externally Validated Models For studies with externally validated prognostic models, further analysis will be done with particular reference to the models. Laupacis et al(23) suggested additional criteria specific to prognostic models in their 1997 paper which was an addition to previous criteria suggested by Wasson et al (24) in 1985,
133
many of which were again identified in the more recent work Hayden et al(5). The factors peculiar to prognostic models from Laupacis et al that are not present in Hayden et al are: Internal validation: Although not explicitly stated and only implied by Laupacis et al, internal validation is important in assessing the performance of the model, although it does not provide information about the model’s performance in another population. This is usually done by splitting the dataset randomly into 2 parts (often 2:1). The model is developed using the first portion and its predictive accuracy is assessed on the second portion. If the available data are limited, the model can be developed on the whole dataset and techniques of data re-use, such as cross validation and bootstrapping, applied to assess performance. (25) External validation: It is important to prospectively validate the model in a group of patients different from the group in which it was derived, and preferably with different clinicians. This examines the generalisability of the model. (25) Sensibility: The evaluation of sensibility relies on judgment rather than statistical methods: Is the model clinically sensible? Clinicians should think that the items in the model are clinically sensible and that no important items are missing. It is difficult to determine which factors are important in the prognosis of seizure outcomes in epilepsy, but this will be determined retrospectively following the systematic review, and then it will subsequently be documented whether these variables were entered into the analysis.(18) Is the model easy to use? This includes factors like time needed to apply the model, and how simple it is to use. Models that require extensive calculations may be less likely to be used than models with simpler scoring schemes. However, this may change as part of a cultural shift involving the increasing surrender of clinical information to the computer and statistical analysis, and this may further facilitate the use of prognostic scores. (26) Probability of outcome vs. Course of action: Laupacis et al(23) are of the opinion that prognostic models that recommend a course of action are more likely to be used compared with those that simply describe the probability of an outcome. Both indices of sensibility will be assessed. Effects of use on clinical practice: This refers to the prospective evaluation of the effect on clinical practice of using the prognostic model in a patient population other than the one in which it was developed and validated to show if physicians and patients are willing to use the model and how its use affect patient behaviour and clinical outcomes. This is best done in randomised controlled trials.(27) REPORT AND SYNTHESIS OF RESULTS / META-ANALYSIS For each study, data will be collected on all of the variables shown to be independently predictive of seizure outcome. The variables included in the prognosis studies will be analysed and reported. Predictive variables consistently found to be independent predictors of an outcome in a range of studies will also be reported. The results of quality assessment will incorporated into the review’s synthesis of the evidence. Information on each of the areas of bias in prognosis studies as identified by Hayden et al(5) will be included in the review synthesis. For example, the evidence of effect would be presented on the basis of studies with low risk for bias associated with study participation, study attrition, prognostic factor measurement, outcome measurement, confounding measurement and account, and analysis. The review synthesis will also include an assessment of evidence of effect based on studies with an overall low risk for any important bias. Therefore studies of acceptable quality for inclusion in the synthesis
134
would at least partly satisfy each of the areas of potential bias i.e. studies from the analysis that are at high risk for any important bias would be omitted from synthesis of results. REFERENCES 1. Hemingway H, Riley R, Altman D. Ten steps towards improving prognosis research. BMJ2009;
339:b4184
2. Moons KG, Royston P, Vergouwe Y, Grobbee DE, Altman DG, Moons KGM, et al. Prognosis and prognostic research: what, why, and how? BMJ2009; 338:b375.
3. Altman D, Lyman G. Methodological challenges in the evaluation of prognostic factors in breast cancer. Breast Cancer Research and Treatment1998;52(1):289-303.
4. Riley RD, Ridley G, Williams K, Altman DG, Hayden J, de Vet HC, et al. Prognosis research: toward evidence-based results and a Cochrane methods group. Journal of Clinical Epidemiology2007 Aug;60(8):863-5; author reply 5-6.
5. Hayden JA, Cote P, Bombardier C, Hayden JA, Cote P, Bombardier C. Evaluation of the quality of prognosis studies in systematic reviews. Annals of Internal Medicine2006 Mar 21;144(6):427-37.
6. Kotsopoulos IA, van Merode T, Kessels FG, de Krom MC, Knottnerus JA, Kotsopoulos IAW, et al. Systematic review and meta-analysis of incidence studies of epilepsy and unprovoked seizures. Epilepsia2002 Nov;43(11):1402-9.
7. Mac TL, Tran DS, Quet F, Odermatt P, Preux PM, Tan CT, et al. Epidemiology, aetiology, and clinical management of epilepsy in Asia: a systematic review. Lancet Neurology2007 Jun;6(6):533-43.
8. Burneo JG, Tellez-Zenteno J, Wiebe S, Burneo JG, Tellez-Zenteno J, Wiebe S. Understanding the burden of epilepsy in Latin America: a systematic review of its prevalence and incidence. Epilepsy Res2005 Aug-Sep;66(1-3):63-74.
9. Benamer HT, Grosset DG, Benamer HTS, Grosset DG. A systematic review of the epidemiology of epilepsy in Arab countries. Epilepsia2009 Oct; 50(10):2301-4.
10. Forsgren L, Beghi E, Oun A, Sillanpaa M. The epidemiology of epilepsy in Europe - a systematic review. Eur J Neurol2005 Apr;12(4):245-53.
11. Preux PM, Druet-Cabanac M, Preux P-M, Druet-Cabanac M. Epidemiology and aetiology of epilepsy in sub-Saharan Africa. Lancet Neurology2005 Jan;4(1):21-31.
12. Ross SD, Estok R, Chopra S, French J. Management of Newly Diagnosed Patients with Epilepsy: A Systematic Review of the Literature. Rockville, MD: Agency for Healthcare Research and Quality (AHRQ) September 2001; http://www.ncbi.nlm.nih.gov/bookshelf/br.fcgi?book=hserta&part=A56819.
13. Kwan P, Sander JW. The natural history of epilepsy: an epidemiological view. Journal of Neurology, Neurosurgery & Psychiatry2004 Oct; 75(10):1376-81.
135
14. Kim LG, Johnson TL, Marson AG, Chadwick DW, group MMS, Kim LG, et al. Prediction of risk of seizure recurrence after a single seizure and early epilepsy: further results from the MESS trial.[Erratum appears in Lancet Neurol. 2006 May;5(5):383]. Lancet Neurology2006 Apr;5(4):317-22.
15. Easterbrook PJ, Berlin JA, Gopalan R, Matthews DR. Publication bias in clinical research. Lancet1991 Apr 13;337(8746):867-72.
16. Shaheen NJ, Crosby MA, Bozymski EM, Sandler RS. Is there publication bias in the reporting of cancer risk in Barrett's esophagus? Gastroenterology2000 Aug;119(2):333-8.
17. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA2000 Apr 19;283(15):2008-12.
18. Counsell C, Dennis M. Systematic review of prognostic models in patients with acute stroke. Cerebrovascular Diseases2001;12(3):159-70.
19. Harrell FE, Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med1996 Feb 28;15(4):361-87.
20. Vittinghoff E, McCulloch CE, Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol2007 Mar 15;165(6):710-8.
21. Wilczynski NL, Haynes RB, Hedges T, Wilczynski NL, Haynes RB. Developing optimal search strategies for detecting clinically sound prognostic studies in MEDLINE: an analytic survey. BMC Medicine2004 Jun 9;2:23.
22. Furlan AD, Irvin E, Bombardier C, Furlan AD, Irvin E, Bombardier C. Limited search strategies were effective in finding relevant nonrandomized studies. Journal of Clinical Epidemiology2006 Dec; 59(12):1303-11.
23. Laupacis A, Sekar N, Stiell IG. Clinical prediction rules. A review and suggested modifications of methodological standards. JAMA1997 Feb 12;277(6):488-94.
24. Wasson JH, Sox HC, Neff RK, Goldman L. Clinical prediction rules. Applications and methodological standards. New England Journal of Medicine1985 Sep 26;313(13):793-9.
25. Altman DG, Vergouwe Y, Royston P, Moons KG, Altman DG, Vergouwe Y, et al. Prognosis and prognostic research: validating a prognostic model. BMJ2009;338:b605.
26. Hemingway H. Prognosis research: Why is Dr. Lydgate still waiting? Journal of Clinical Epidemiology2006; 59(12):1229-38.
27. Moons KG, Altman DG, Vergouwe Y, Royston P, Moons KGM, Altman DG, et al. Prognosis and prognostic research: application and impact of prognostic models in clinical practice. BMJ2009; 338:b606.
136
7.2 Appendix II: LIST OF CITATIONS This is the list of citations from screening endnote library (MEDLINE & EMBASE) and references of eligible publications LEGENDS
From MEDLINE Only From EMBASE Only From EMBASE & MEDLINE From References Yes - Eligible | No - Not Eligible | MVA - Multivariate Analysis | RO - Relevant Outcome | NMA - No Multivariate Analysis | NRO - No Relevant Outcome EVS - External Validation Study | CC - Case Control Study | R - Restricted Epilepsy Population | <100 - Fewer than 100 Patients
CITATIONS
ELIGIBILITY
Abduljabbar M, Ogunniyi A, Daif AK, Al-Tahan A, Al-Bunyan M, Al-Rajeh S. Epilepsy classification and factors associated with control in Saudi adult patients. Seizure1998 Dec;7(6):501-4. Yes (RO, MVA)
Akhondian J, Heydarian F, Jafari SA, Akhondian J, Heydarian F, Jafari S-A. Predictive factors of pediatric intractable seizures. Arch Iran Med2006 Jul;9(3):236-9. No (NMA) CC
Albright P, Bruni J. Reduction of polypharmacy in epileptic patients. Arch Neurol1985 Aug;42(8):797-9. No (NRO)
Altunbasak S, Incecik F, Herguner O, Refik Burgut H, Altunbasak S, Incecik F, et al. Prognosis of patients with seizures occurring in the first 2 years. J Child Neurol2007 Mar;22(3):307-13. No (RO, MVA) <100
Annegers JF, Hauser WA, Elveback LR. Remission of seizures and relapse in patients with epilepsy. Epilepsia1979 Dec;20(6):729-37. No (NMA)
Anonymous. Prognosis of epilepsy in newly referred patients: a multicenter prospective study of the effects of monotherapy on the long-term course of epilepsy. Epilepsia1992 Jan-Feb;33(1):45-51. Yes (RO, MVA)
Arts WF, Brouwer OF, Peters AC, Stroink H, Peeters EA, Schmitz PI, et al. Course and prognosis of childhood epilepsy: 5-year follow-up Brain. 2004 Aug;127(Pt 8):1774-84. Yes (RO, MVA)
Arts WF, Geerts AT, Brouwer OF, Boudewyn Peters AC, Stroink H, van Donselaar CA. The early prognosis of epilepsy in childhood: the prediction of a poor outcome. Epilepsia. 1999 Jun;40(6):726-34. Yes (RO, MVA)
Attarian H, Vahle V, Carter J, Hykes E, Gilliam F. Relationship between depression and intractability of seizures. Epilepsy Behav2003 Jun;4(3):298-301. No (NMA)
Baker GA, Gagnon D, McNulty P. The relationship between seizure frequency, seizure type and quality of life: findings from three European countries. Epilepsy Res1998 May;30(3):231-40. No (NRO)
Balan V, Garg AR, Vaidyalingam M, Vyas JN, Pasha CK, Mani KS. Epilepsy. II. Follow-up study of 1010 cases of epilepsy. Neurol India1968 Apr-Jun;16(2):62-5. No (NMA)
Banu SH, Khan NZ, Hossain M, Jahan A, Parveen M, Rahman N, et al. Profile of childhood epilepsy in Bangladesh. Dev Med Child Neurol. 2003 Jul;45(7):477-82. Yes (RO, MVA)
Battaglia D, Rando T, Deodato F, Bruccini G, Baglio G, et al. Epileptic disorders with onset in the first year of life: neurological & cognitive outcome. Europ J Paediatr Neurol1999;3(3):95-103. No (NMA)
Bauer J, Buchmuller L, Reuber M, Burr W. Which patients become seizure free with antiepileptic drugs? An observational study in 821 patients with epilepsy. Acta Neurol Scand2008 Jan;117(1):55-9. No (NRO)
Bautista RE, Glen ET, Shetty NK, Wludyka P, Bautista RED, Glen ET, et al. The association between health literacy and outcomes of care among epilepsy patients. Seizure2009 Jul;18(6):400-4. No (NRO)
Beghi E, Tognoni G. Prognosis of epilepsy in newly referred patients: a multicenter prospective study. Collaborative Group for the Study of Epilepsy. Epilepsia. 1988 May-Jun;29(3):236-43. Yes (RO, MVA)
Berg AT, Levy SR, Novotny EJ, Shinnar S. Predictors of intractable epilepsy in childhood: a case-control study. Epilepsia1996 Jan;37(1):24-30. Yes (RO, MVA)
Berg AT, Levy SR, Testa FM, D'Souza R, Berg AT, Levy SR, et al. Remission of epilepsy after two drug failures in children: a prospective study. Ann Neurol2009 May;65(5):510-9. No (NRO)
Berg AT, Shinnar S, Levy SR, Testa FM, Smith-Rapaport S, Beckerman B. Early development of intractable epilepsy in children: a prospective study. Neurology. 2001 Jun 12;56(11):1445-52. Yes (RO, MVA)
137
Berg AT, Shinnar S, Levy SR, Testa FM, Smith-Rapaport S, Beckerman B, et al. Two-year remission and subsequent relapse in children with newly diagnosed epilepsy. Epilepsia Yes (RO, MVA)
Berg AT, Shinnar S, Levy SR, Testa FM, Smith-Rapaport S, Beckerman B, et al. Defining early seizure outcomes in pediatric epilepsy: the good, bad and in-between. Epilepsy Res2001 Jan;43(1):75-84. No (NMA)
Berg AT, Vickrey BG, Testa FM, Levy SR, Shinnar S, DiMario F, et al. How long does it take for epilepsy to become intractable? A prospective investigation. Ann Neurol. No (NMA)
Bonnett L, Tudur-Smith C, Williamson P, Marson T. Prognostic factors for 12-month remission. Epilepsia2009;50:110. No (NMA)
Brodbeck V, Jansen V, Fietzek U, Muehe C, Weber G, Heinen F, et al. Long-term profile of lamotrigine in 119 children with epilepsy. Europ J Paediatr Neurol. 2006 May;10(3):135-41. No (NRO)
Braathen G, Andersson T, Gylje H, Melander H, Naglo AS, et al. Comparison between one and three years of treatment in uncomplicated childhood epilepsy. I. Epilepsia1996 Sep;37(9):822-32. No (NRO)
Braathen G, Melander H. Early discontinuation of treatment in children with uncomplicated epilepsy: a prospective study with a model for prediction of outcome. Epilepsia1997 May;38(5):561-9. No (NRO)
Brorson LO, Wranne L. Long-term prognosis in childhood epilepsy: survival and seizure prognosis. Epilepsia1987 Jul-Aug;28(4):324-30. No (NMA)
Buchanan N. Clobazam in the treatment of epilepsy: prospective follow-up to 8 years. J R Soc Med1993 Jul;86(7):378-80. No (NRO)
Burgess RJ, Drake ME, Jr., Paulson GW. Interictal EEG prediction of response to antiepileptic drug monotherapy. Clin Electroencephalogr1985 Jul;16(3):131-5. No (NMA)
Camfield C, Camfield P, Gordon K, Smith B, Dooley J. Outcome of childhood epilepsy: a scoring system for those treated with medication. J Pediatr. 1993 Jun;122(6):861-8. Yes (RO, MVA)
Camfield C, Camfield P, Smith B, Gordon K, Dooley J. Biologic factors as predictors of social outcome of epilepsy in intellectually normal children. J Pediatr1993 Jun;122(6):869-73. No (NRO)
Camfield P, Camfield C, Camfield P, Camfield C. The frequency of intractable seizures after stopping AEDs in seizure-free children with epilepsy. Neurology2005 Mar 22;64(6):973-5. No (NRO)
Camfield P, Camfield C, Camfield P, Camfield C. Long-term prognosis for symptomatic (secondarily) generalized epilepsies: a population-based study. Epilepsia. 2007 Jun;48(6):1128-32. No (NRO)
Camfield P, Camfield C, Smith S, Dooley J, Smith E, Camfield P, et al. Long-term outcome is unchanged by antiepileptic drug treatment after a first seizure. Epilepsia2002 Jun;43(6):662-3. No (NRO)
Camfield PR, Camfield CS, Gordon K, Dooley JM. If a first antiepileptic drug fails to control a child's epilepsy, what are the chances of success with the next drug? J Pediatr1997 Dec;131(6):821-4. No (NMA)
Carpay HA, Arts WF, Geerts AT, Stroink H, Brouwer OF, Boudewyn Peters AC, et al. Epilepsy in childhood: an audit of clinical practice. Arch Neurol1998 May;55(5):668-73. No (NMA)
Casetta I, Granieri E, Monetti VC, Gilli G, Tola MR, Paolino E, et al. Early predictors of intractability in childhood epilepsy. Acta Neurol Scand. 1999 Jun;99(6):329-33. Yes (RO, MVA)
Casetta I, Granieri E, Monetti VC, Tola MR, Paolino E, Malagu S, et al. Prognosis of childhood epilepsy: a community-based study in Copparo, Italy. Neuroepidemiology. 1997;16(1):22-8. No (R)
Cavazzuti GB, Cappella L, Nalin A. Longitudinal study of epileptiform EEG patterns in normal children. Epilepsia1980 Feb;21(1):43-55. No (NRO)
Chakova L. Studies on the frequency and clinical manifestations of epilepsy in infancy and early childhood. Folia Med (Plovdiv)1996;38(2):75-80. No (NMA)
Chawla S, Aneja S, Kashyap R, Mallika V, Chawla S, Aneja S, et al. Etiology and clinical predictors of intractable epilepsy. Pediatr Neurol2002 Sep;27(3):186-91. Yes (RO, MVA)
Chevrie JJ, Aicardi J. Convulsive disorders in the first year of life: neurological and mental outcome and mortality. Epilepsia1978 Feb;19(1):67-74. No (NRO)
Clemens B. Timing discontinuation of antiepileptic treatment in childhood epilepsies--the role of the sleep deprivation EEG: a preliminary study. Jpn J Psychiatry Neurol1989 Mar;43(1):85-8. No (NRO)
Cockerell OC, Eckle I, Goodridge DM, Sander JW, Shorvon SD. Epilepsy in a population of 6000 patients.J Neurol Neurosurg Psychiatry1995 May;58(5):570-6. No (NRO)
Cockerell OC, Johnson AL, Sander JW, Hart YM, Shorvon SD. Remission of epilepsy: results from the National General Practice Study of Epilepsy. Lancet. 1995 Jul 15;346(8968):140-4. No (NMA)
Cockerell OC, Johnson AL, Sander JW, Shorvon SD. Prognosis of epilepsy: a review and further analysis of the first nine years of the BNGPSE. Epilepsia.. 1997 Jan;38(1):31-46. No (NMA)
Czochanska J, Langner-Tyszka B, Losiowski Z, Schmidt-Sidor B. Children who develop epilepsy in the first year of life: a prospective study. Dev Med Child Neurol1994 Apr;36(4):345-50. No (NRO)
Czochanska J, Losiowski Z, Langner-Tyszka B, Bielicka-Cymerman J, et al. Intractable epilepsy in children who develop epilepsy in the first decade of life. Mater Med Pol1996 Oct-Dec;28(4):133-7. No (NMA)
Danesi MA. Prognosis of seizures in medically-treated adolescent and adult Nigerian epileptics. Trop Geogr Med1983 Dec;35(4):395-9. No (NMA)
Datta AN, Wirrell EC. Prognosis of seizures occurring in the first year. Pediatr Neurol2000 May;22(5):386-91. No (NMA)
Del Felice A, Beghi E, Boero G, La Neve A, Bogliun G, De Palo A, et al. Early versus late remission in a cohort of patients with newly diagnosed epilepsy. Epilepsia;51(1):37-42. Yes (RO, MVA)
Di Mascio R, Beghi E, Sasanelli F, Tognoni G. Early prognosis of epilepsy. Effects of treatment in the first follow-up year. Ital J Neurol Sci. 1986 Aug;7(4):421-9. No (NMA)
Dudley RW, Penney SJ, Buckley DJ, Dudley RWR, Penney SJ, Buckley DJ. First-drug treatment failures in children newly diagnosed with epilepsy. Pediatr Neurol2009 Feb;40(2):71-7. No (NMA)
Ekholm E, Niemineva K. On convulsions in early childhood and their prognosis; an investigation with follow-up examinations. Acta Paediatr1950;39(6):481-501. No (NRO)
138
Elwes RD, Johnson AL, Reynolds EH. The course of untreated epilepsy. Bmj1988 Oct 15;297(6654):948-50. No (NRO)
Elwes RD, Johnson AL, Shorvon SD, Reynolds EH. The prognosis for seizure control in newly diagnosed epilepsy. N Engl J Med. 1984 Oct 11;311(15):944-7. No (NMA)
Farmer PJ, Placencia M, Jumbo L, Sander JW, Shorvon SD. Effects of epilepsy on daily functioning in northern Ecuadorn. Neuroepidemiology1992;11(4-6):180-9. No (NRO)
Fukushima Y. A study on long-term prognosis of epilepsy. Folia Psychiatr Neurol Jpn1977;31(3):369-74. No (NMA)
Gauffin H, Raty L, Soderfeldt B, Gauffin H, Raty L, Soderfeldt B. Medical outcome in epilepsy patients of young adulthood--a 5-year follow-up study. Seizure2009 May;18(4):293-7. No (NMA)
Geelhoed M, Boerrigter AO, Camfield P, Geerts AT, Arts W, Smith B, et al. The accuracy of outcome prediction models for childhood-onset epilepsy. Epilepsia. 2005 Sep;46(9):1526-32. Yes (RO, MVA, EVS)
Geerts AT, Arts WF, Brouwer OF, Peters AC, Peeters EA, Stroink H, et al. Validation of two prognostic models predicting outcome at two years after diagnosis .Epilepsia2006;47(6):960-5. Yes (EVS)
Goodridge DM, Shorvon SD. Epileptic seizures in a population of 6000. II: Treatment and prognosis. Br Med J (Clin Res Ed).. 1983 Sep 3;287(6393):645-7. No (NMA)
Gururaj A, Sztriha L, Hertecant J, Eapen V, Gururaj A, Sztriha L, et al. Clinical predictors of intractable childhood epilepsy. J Psychosom Res2006 Sep;61(3):343-7. No (NMA) CC
Harrison RM, Taylor DC. Childhood seizures: a 25-year follow up. Social and medical prognosis. Lancet.. 1976 May 1;1(7966):948-51. No (NRO)
Hasegawa H, Kanner AM. Seizure control after chronic pharmacotherapy in epileptic disorders beginning after 40 years of age. Clin Neuropharmacol1995 Feb;18(1):13-22. No (NMA)
Hauser E, Freilinger M, Seidl R, Groh C. Prognosis of childhood epilepsy in newly referred patients. J Child Neurol1996 May;11(3):201-4. No (NMA)
Holt-Seitz A, Wirrell EC, Sundaram MB. Seizures in the elderly: etiology and prognosis. Can J Neurol Sci1999 May;26(2):110-4. No (NMA)
Hosokawa K, Kugoh T. Multidimensional clinical study of epileptics under long-term follow-up. Folia Psychiatr Neurol Jpn1977;31(3):359-67. No (NRO)
Hosokawa K, Kugoh T. Prognosis in patients with epilepsy--special reference to change in types of seizure and ingestion of prescribed drugs. Folia Psychiatr Neurol Jpn1978;32(3):447-8. No (NMA)
Hui ACF, Wong A, Wong HC, Man BL, Au-Yeung KM, Wong KS. Refractory epilepsy in a Chinese population. Clinical Neurology and Neurosurgery2007;109(8):672-5. Yes (RO, MVA)
Hull RP, Haerer AF. Follow-up of epileptic outpatients. South Med J1973 Mar;66(3):292-6. No (NMA)
Hyllested K, Pakkenberg H. Prognosis in epilepsy og late onset. Neurology1963 Aug;13:641-4. No (NMA)
Jain S, Maheshwari MC. A prolonged prospective follow-up study of 306 epileptic patients in New Delhi. Acta Neurol Scand1991 Dec;84(6):471-4. No (NMA)
Jalava M, Sillanpaa M. Concurrent illnesses in adults with childhood-onset epilepsy: a population-based 35-year follow-up study. Epilepsia. 1996 Dec;37(12):1155-63. No (NRO)
Jalava M, Sillanpaa M. Physical activity, health-related fitness, and health experience in adults with childhood-onset epilepsy: a controlled study. Epilepsia1997 Apr;38(4):424-9. No (NRO)
Jalava M, Sillanpaa M, Camfield C, Camfield P. Social adjustment and competence 35 years after onset of childhood epilepsy: a prospective controlled study. Epilepsia1997 Jun;38(6):708-15. No (NRO)
Jilek-Aall L, Rwiza HT. Prognosis of epilepsy in a rural African community: a 30-year follow-up of 164 patients in an outpatient clinic in rural Tanzania. Epilepsia1992 Jul-Aug;33(4):645-50. No (NRO)
Kalita J, Vajpeyee A, Misra UK. Predictors of one-year seizure remission--a clinicoradiological and electroencephalographic study. Electromyogr Clin Neurophysiol2005 Apr-May;45(3):161-6. No (NMA)
Kim LG, Johnson TL, Marson AG, Chadwick DW, group MMS, Kim LG, et al. Prediction of risk of seizure recurrence after a single seizure and early epilepsy Lancet neurol2006 Apr;5(4):317-22. Yes (RO, MVA)
Kiorboe E. The prognosis of epilepsy. Acta Psychiatr Scand Suppl1961;36(150):166-78. No (NMA)
Kitagawa T. A clinical and electroencephalographical follow-up study for more than 10 years in patients with epilepsy. Folia Psychiatr Neurol Jpn1981;35(3):333-42. No (NMA)
Ko TS, Holmes GL. EEG and clinical predictors of medically intractable childhood epilepsy. Clin Neurophysiol1999 Jul;110(7):1245-51. Yes (RO, MVA)
Kochen S, Melcon MO. Prognosis of epilepsy in a community-based study: 8 years of follow-up in an Argentine community. Acta Neurol Scand. 2005 Dec;112(6):370-4. No (NMA)
Koprowski C, Clancy R. Clinical predictors of computed tomography in epileptic children. J Child Neurol. 1986 Apr;1(2):145-8. No (NRO)
Kotsopoulos I, de Krom M, Kessels F, Lodder J, Troost J, Twellaar M, et al. Incidence of epilepsy and predictive factors of epileptic and non-epileptic seizures. Seizure. 2005 Apr;14(3):175-82. No (NRO)
Kramer U, Nevo Y, Neufeld MY, Fatal A, Leitner Y, Harel S. Epidemiology of epilepsy in childhood: a cohort of 440 consecutive patients. Pediatr Neurol1998 Jan;18(1):46-50. No (NRO)
Kuhl V, Kiorboe E, Lund M. The prognosis of epilepsy with special reference to traffic security. Epilepsia1967 Sep;8(3):195-209. No (NMA)
Kurtz Z, Tookey P, Ross E. Epilepsy in young people: 23 year follow up of the British national child development study. BMJ 1998 Jan 31;316(7128):339-42. No (NRO)
Kwan P, Brodie MJ. Epilepsy after the first drug fails: substitution or add-on? Seizure2000 Oct;9(7):464-8. No (NRO)
139
Kwan P, Brodie MJ. Early identification of refractory epilepsy. N Engl J Med2000 Feb 3;342(5):314-9. No (NMA)
Kwong KL, Chak WK, Wong SN, So KT. Epidemiology of childhood epilepsy in a cohort of 309 Chinese children. Pediatr Neurol2001 Apr;24(4):276-82. No (NMA)
Kwong KL, Sung WY, Wong SN, So KT, Kwong KL, Sung WY, et al. Early predictors of medical intractability in childhood epilepsy. Pediatr Neurol2003 Jul;29(1):46-52. Yes (RO, MVA)
Lehtinen LO. Prognosis of epilepsy. A preliminary report. Acta Neurologica Scandinavica Supplementum1965;13 Pt 2:517-20. No (NRO)
Lehtovaara R, Elomaa E, Bardy A. A follow-up study of 880 adult ambulatory epileptics. Monogr Neural Sci1980;5:273-6. No (NMA)
Lhatoo SD, Sander JW, Shorvon SD. The dynamics of drug treatment in epilepsy: an observational study diagnosed epilepsy. J Neurol Neurosurg Psychiatry2001 Nov;71(5):632-7. No (NMA)
Lindsten H, Stenlund H, Forsgren L. Seizure recurrence in adults after a newly diagnosed unprovoked epileptic seizure. Acta Neurol Scand2001 Oct;104(4):202-7. Yes (RO, MVA)
Lindsten H, Stenlund H, Forsgren L. Remission of seizures in a population-based adult cohort with a newly diagnosed unprovoked epileptic seizure. Epilepsia. 2001 Aug;42(8):1025-30. No (NMA)
Lohani S, Devkota UP, Rajbhandari H. Predictors of unfavourable seizure outcome in patients with epilepsy in Nepal. Canadian Journal of Neurological Sciences;37(1):76-80. Yes (RO, MVA)
Loiseau P, Pestre M, Dartigues JF. Symptomatology and prognosis in adolescent epilepsies (a study of 1033 cases). Epilepsy Res1987 Sep;1(5):290-6. No (NMA)
Lossius MI, Stavem K, Gjerstad L. Predictors for recurrence of epileptic seizures in a general epilepsy population. Seizure1999 Dec;8(8):476-9. Yes (RO, MVA)
Luhdorf K, Jensen LK, Plesner AM. Epilepsy in the elderly: prognosis. Acta Neurol Scand1986 Nov;74(5):409-15. No (NMA)
MacDonald BK, Johnson AL, Goodridge DM, Cockerell OC, Sander JW, Shorvon SD. Factors predicting prognosis of epilepsy after presentation with seizures. Ann Neurol. 2000 Dec;48(6):833-41. Yes (RO, MVA)
Malik MA, Hamid MH, Ahmed TM, Ali Q, Malik MA, Hamid MH, et al. Predictors of intractable childhood epilepsy. J Coll Physicians Surg Pak2008 Mar;18(3):158-62. No (NMA) CC
Mohanraj R, Brodie MJ, Mohanraj R, Brodie MJ. Pharmacological outcomes in newly diagnosed epilepsy. Epilepsy Behav2005 May;6(3):382-7. No (NRO)
Mondkar VP, Manikal MD, Kohiyar FN, Bharucha EP. Epilepsy. (A follow-up study of 170 cases). J Postgrad Med1969 Oct;15(4):165-75. No (NMA)
Morikawa T, Ishihara O, Kakegawa N, Seino M, Wada T. A retrospective study on the prognosis of aged patients with epilepsy. Folia Psychiatr Neurol Jpn1977;31(3):375-81. No (NRO)
Nicoletti A, Sofia V, Vitale G, Bonelli SI, Bejarano V, Bartalesi F, et al. Natural history and mortality of chronic epilepsy in an untreated population of rural Bolivia. Epilepsia. 2009 Oct;50(10):2199-206. No (NMA)
Ohtahara S, Yamatogi Y, Ohtsuka Y, Oka E, Kanda S. Prognosis in childhood epilepsy: a prospective follow-up study. Folia Psychiatr Neurol Jpn1977;31(3):301-13. No (NMA)
Ohtsuka Y, Ogino T, Murakami N, Amano R, Enoki H, Yamatogi Y, et al. Refractory epilepsy in infancy and childhood--a prospective follow-up study. Jpn J Psychiatry Neurol1987 Sep;41(3):440-3. No (NMA) CC
Oka E, Yamatogi Y, Ohtsuka Y, Ohtahara S. Clinical course and prognosis of childhood epilepsy. Acta Paediatr Jpn1989 Jun;31(3):259-66. No (NMA)
Okuma T. Prognosis of epilepsy: a preliminary report of a multi-institutional study in Japan. Folia Psychiatr Neurol Jpn1977;31(3):289-99. No (NMA)
Okuma T, Kumashiro H. Prognosis of epilepsy: the second interim report of a multi-institutional study in Japan. Folia Psychiatr Neurol Jpn1978;32(3):421-31. No (NMA)
Okuma T, Kumashiro H. Natural history and prognosis of epilepsy: report of a multi-institutional study in Japan. Epilepsia. 1981 Feb;22(1):35-53. No (NMA)
Oskoui M, Webster RI, Zhang X, Shevell MI, Oskoui M, Webster RI, et al. Factors predictive of outcome in childhood epilepsy. J Child Neurol, 2005 Nov;20(11):898-904. Yes (RO, MVA)
Placencia M, Sander JW, Roman M, Madera A, et al. The characteristics of epilepsy in a largely untreated population in rural Ecuador. J Neurol Neurosurg Psychiatry. 1994 Mar;57(3):320-5. No (NMA)
Ramos-Lizana J, Aguilera-Lopez P, Aguirre-Rodriguez J, Cassinello-Garcia E. Early prediction of refractory epilepsy in childhood. Seizure2009 Jul;18(6):412-6. Yes (RO, MVA)
Ranheim B, Dietrichson P, Williamsen R. Prognosis in epilepsy. II. Further report of an investigation. Acta Neurologica Scandinavica Supplementum1965;13 Pt 2:497-507. No (NRO)
Rankin RM. Prognosis of epilepsy in children. Northwest Med1972 Jun;71(6):455-9. No (NMA)
Rantala H, Ingalsuo H. Occurrence and outcome of epilepsy in children younger than 2 years. J Pediatr. [Research Support, Non-U.S. Gov't]. 1999 Dec;135(6):761-4. No (RO, MVA) <100
Rodin EA, Rhodes RJ, Velarde NN. The prognosis for patients with epilepsy. J Occup Med1965 Nov;7(11):560-3. No (RO, MVA) <100
Sawhney IM, Singh A, Kaur P, Suri G, Chopra JS. A case control study and one year follow-up of registered epilepsy cases in a resettlement colony of North India. J Neurol Sci1999 May 1;165(1):31-5. No (NMA) CC
Semah F, Picot MC, Adam C, Broglin D, Arzimanoglou A, Bazin B, et al. Is the underlying cause of epilepsy a major prognostic factor for recurrence? Neurology1998 Nov;51(5):1256-62. No (R)
Shackleton DP, Kasteleijn-Nolst Trenite DG, de Craen AJ, Vandenbroucke JP, et al. Living with epilepsy: long-term prognosis and psychosocial outcomes. Neurology. 2003 Jul 8;61(1):64-70. No (NRO)
Shafer SQ, Hauser WA, Annegers JF, Klass DW. EEG and other early predictors of epilepsy remission: a community study. Epilepsia1988 Sep-Oct;29(5):590-600. Yes (RO, MVA)
140
Shinnar S, O'Dell C, Berg AT. Distribution of epilepsy syndromes in a cohort of children prospectively monitored from the time of their first unprovoked seizure. Epilepsia1999 Oct;40(10):1378-83. No (NMA)
Shinnar S, Berg AT, O'Dell C, Newstein D, at al. Predictors of multiple seizures in a cohort of children followed from the time of their first unprovoked seizure. Ann Neurol2000 Aug;48(2):140-7. Yes (RO, MVA)
Shorvon SD, Reynolds EH. Early prognosis of epilepsy. Br Med J 1982 Dec 11;285(6356):1699-701. No (NMA)
Sillanpaa M. Children with epilepsy as adults: outcome after 30 years of follow-up. Acta Paediatr Scand Suppl1990;368:1-78. Yes (RO, MVA)
Sillanpaa M. Remission of seizures and predictors of intractability in long-term follow-up. Epilepsia1993 Sep-Oct;34(5):930-6. Yes (RO, MVA)
Sillanpaa M, Camfield P, Camfield C. Predicting long-term outcome of childhood epilepsy in Nova Scotia, and Turku. Validation of a scoring system. Arch Neurol1995 Jun;52(6):589-92. Yes (EVS)
Sillanpaa M, Jalava M, Kaleva O, Shinnar S. Long-term prognosis of seizures with onset in childhood. N Engl J Med. Yes (RO, MVA)
Sillanpaa M, Jalava M, Shinnar S. Epilepsy syndromes in patients with childhood-onset seizures in Finland. Pediatr Neurol1999 Aug;21(2):533-7. No (NMA)
Sillanpaa M, Schmidt D, Sillanpaa M, Schmidt D. Natural history of treated childhood-onset epilepsy: prospective, long-term population-based study. Brain2006 Mar;129(Pt 3):617-24. Yes (RO, MVA)
Sillanpaa M, Schmidt D, Sillanpaa M, Schmidt D. Early seizure frequency and aetiology predict long-term medical outcome in childhood-onset epilepsy. Brain. 2009 Apr;132(Pt 4):989-98. Yes (RO, MVA)
Sofijanov NG. Clinical evolution and prognosis of childhood epilepsies. Epilepsia1982 Feb;23(1):61-9. No (NMA)
Stephen LJ, Kelly K, Mohanraj R, Brodie MJ, Stephen LJ, Kelly K, et al. Pharmacological outcomes in older people with newly diagnosed epilepsy. Epilepsy Behav2006 Mar;8(2):434-7. No (NMA)
Strobos RR. Prognosis in convulsive disorders. Arch Neurol1959 Aug;1:216-25. No (NMA)
Suzuki M, Suzuki Y, Mizuno T, Konishi Y, Mizuno Y. Long-term prognosis of epilepsy in children--a follow-up report beyond 18 years of age. Folia Psychiatr Neurol Jpn1976;30(3):307-13. No (NMA)
Suzuki N, Seki T, Yamawaki H, Kimiya S, Maezawa M, Tachibana Y, et al. Long-term prognosis of childhood epilepsy--follow-up until adulthood . Jpn J Psychiatry Neurol1987 Sep;41(3):437-9. No (NMA)
Thapar A, Roland M, Harold G, Thapar A, Roland M, Harold G. Do depression symptoms predict seizure frequency--or vice versa? J Psychosom Res2005 Nov;59(5):269-74. No (NRO)
Tsuboi T. Seizures of childhood. A population-based and clinic-based study. Acta Neurologica Scandinavica Supplementum. 1986;110:1-237. No (NMA)
Udani VP, Dharnidharka V, Nair A, Oka M. Difficult to control epilepsy in childhood--a long term study of 123 cases. Indian Pediatr1993 Oct;30(10):1199-206. No (NRO)
Vanderlinden L, Lagae LG, Vanderlinden L, Lagae LG. Clinical predictors for outcome in infants with epilepsy. Pediatr Neurol2004 Jul;31(1):52-5. No (NMA)
Wada K, Kiryu K, Kawata Y, Chiba T, Mizuno K, Okada M, et al. Prognosis and clinical features of intractable epilepsy: a prospective study. Psychiatry Clin Neurosci1997 Aug;51(4):233-5. No (NRO)
Wakamoto H, Nagao H, Hayashi M, Morimoto T. Long-term medical, educational, and social prognoses of childhood-onset epilepsy. Brain Dev2000 Jun;22(4):246-55. No (NMA)
Watts AE. The natural history of untreated epilepsy in a rural community in Africa. Epilepsia1992 May-Jun;33(3):464-8. No (NRO)
Williamsen R, Ranheim B, Dietrichson P. Prognosis in epilepsy. Preliminary report of an investigation. Acta Neurologica Scandinavica Supplementum1963;39(4):SUPPL4:349-63. No (NMA)
Wilson WP, Stewart LF, Parker JB, Jr. Epilepsy: some factors influencing the prognosis and treatment. Tex State J Med1960 Jan;56:31-3. No (NRO)
Wu LW, Kugimiya S, Numata Y, Wada T. Prognostic factors of primary generalized epilepsy: a reappraisal of 96 cases in terminal remission. Folia Psychiatr Neurol Jpn. 1985;39(2):139-45. No (NMA)
Yamada H, Yoshida H, Ninomiya H. A five-year follow-up study of 66 epileptics. Folia Psychiatr Neurol Jpn1977;31(3):339-45. No (NMA)
Yamada H, Yoshida H, Ninomiya H, Kato Y. A comparison of follow-up studies on epilepsy--five year and ten-year prognoses. Folia Psychiatr Neurol Jpn. 1978;32(3):451-2. No (NMA)
Yamada H, Yoshida H, Ninomiya H, Kato Y. A 10-year follow-up study of 97 epileptics. Folia Psychiatr Neurol Jpn1979;33(2):172-82. No (NMA)
Yamada H, Yoshida H, Ninomiya H, Kato Y. The social prognosis of epileptics--a 10-year follow-up study. Folia Psychiatr Neurol Jpn1980;34(3):306-7. No (NRO)
Yamamoto Y, Kuwahara H, Miyauchi T, Hosaka H, Yokoi S. Follow-up survey on 131 dropout cases of epilepsy. Folia Psychiatr Neurol Jpn1980;34(3):299. No (NRO)
141
7.3 Appendix III: DATA EXTRACTION AND QUALITY APPRAISAL FORM This contains the three versions of the data extraction and quality appraisal form, from the first two pilots (Versions 1 and 2) to the final version (Version 3)
Data Extraction and Quality Appraisal Form Version 1 Data Extraction First Author, Year, Title & Journal: _________________________________________________________________________________
Study Design Prospective Retrospective Mixed Unclear Type of Study Randomised Controlled Trial Cohort Study Nested Case Control Study Case Control Study
Study Population Characteristics Gender Male Female Both
Age Bracket ________________
Remark Children Adolescents Adults Setting Hospital Population Both
Follow up Population Initial________, Final_________, %__________ Not Stated/ Unclear
Length of Reported Follow up Minimal_________, Average_________, Median_________, Maximum__________ Not Stated/ Unclear
Outcome Measure Seizure
Remission – terminal remission, longest remission Seizure status at specified points within follow-up timeline Time to event – remission, recurrence of seizures, intractability
Time to recurrence of seizures Intractability Refractoriness
Epilepsy Subtypes Included/Excluded
All Included/ Unselected Undetermined/ Not stated / Not sure
Method of Multivariate Analysis Logistic Regression Model Cox Proportional Hazards Model
Treatment Details of Cohort _____________________________________________________________________________________________
Independent Predictors (with Regression Coefficient, Odds Ratio or Hazard Ratio and 95% CI)
142
Quality Appraisal for All Studies Study participation The source population or population of interest is adequately described for key characteristics Yes Partly No Unsure
The sampling frame and recruitment are adequately described Yes Partly No Unsure Inclusion and exclusion criteria are adequately described Yes Partly No Unsure There is adequate participation in the study by eligible individuals Yes Partly No Unsure
The baseline study sample (or cohort) is adequately described for key characteristics Yes Partly No Unsure Study attrition Response rate (proportion of study sample completing the study and providing outcome data) is adequate (>90%) Yes Partly No Unsure
Information on participants who dropped out of the study are described Yes Partly No Unsure Reasons for loss to follow-up are provided Yes Partly No Unsure Participants lost to follow-up are adequately described for key characteristics Yes Partly No Unsure
There are no important differences between those who completed the study and those who did not Yes Partly No Unsure Prognostic factor measurement
The prognostic factors measured are clearly defined or described where necessary Yes Partly No Unsure Continuous variables are reported or appropriate (i.e. not data-dependent) cut-points are used Yes Partly No Unsure The prognostic factor measure and method are adequately valid and reliable to limit misclassification bias Yes Partly No Unsure
The method and setting of measurement are the same or similar for all study participants Yes Partly No Unsure Adequate proportion of the study sample has complete data for prognostic factors Yes Partly No Unsure Appropriate methods are used if imputation is used for missing prognostic factor data Yes Partly No Unsure
Outcome measurement A clear definition of the outcome of interest is provided, including duration of follow-up and level and extent of the outcome construct Yes Partly No Unsure The outcome measure and method used are adequately valid and reliable to limit misclassification bias Yes Partly No Unsure
The method and setting of measurement are the same or similar for all study participants Yes Partly No Unsure
Confounding measurement and account
Important confounders are measured Yes Partly No Unsure Clear definitions of the important confounders measured are provided Yes Partly No Unsure Measurement of important confounders is adequately valid and reliable Yes Partly No Unsure
The method and setting of confounding measurement are the same for all study participants. Yes Partly No Unsure Appropriate methods are used if imputation is used for missing confounder data Yes Partly No Unsure Important potential confounders are accounted for in the study design (e.g. matching, stratification, or initial assembly of comparable groups) Yes Partly No Unsure
Important potential confounders are accounted/adjusted for in the analysis Yes Partly No Unsure Analysis There is sufficient presentation of data to assess the adequacy of the analysis Yes Partly No Unsure
The strategy for model building (i.e. inclusion of variables) is appropriate and is based on a conceptual framework or model Yes Partly No Unsure The selected model is adequate for the design of the study Yes Partly No Unsure There is no selective reporting of results Yes Partly No Unsure
143
Data Extraction and Quality Appraisal Form Version 2 Data Extraction First Author, Year, Title & Journal: ___________________________________________________________________________________ Study Design
Prospective Retrospective Mixed Unclear Type of Study
Randomised Controlled Trial Non-Randomised Controlled Trial Cohort Study Nested Case Control Study Case Control Study
Study Setting
Hospital Population
Study Recruitment Period From___________________ To________________________ (n= )
Age Bracket ________________, Mean ___________ ± _________SD
Follow up Initial Population ________, Final_________, %__________ Not Stated/ Unclear
Length of Reported Follow up Minimal_________, Mean_________, Median_________, Maximum__________ Not Stated/ Unclear
Outcome More Information Seizure Outcome Measure
Remission
Time to Event Intractability/ Refractoriness Status at specified points within follow-up timeline
Others ______________________________________ Proportion of Patients with Outcome ______________________%___________ Not Stated/ Unclear
Independent Predictors of Seizure Outcome (with Regression Coefficient, Odds Ratio or Hazard Ratio and 95% CI)
1. 2. 3.
4.
5. 6.
144
Epilepsy Subtypes Included/Excluded More Information All Included - Unselected
Not stated - Not sure Selected
Method of Multivariate Analysis More Information Logistic Regression Model Cox Proportional Hazards Model
Number of Predictor Variables entered into Model__________________________ Not Stated/ Unclear
ILAE Reference Standards More Information
Definition of Epilepsy Yes Partly No Unsure Classification of Seizures Yes Partly No Unsure
Classification by Aetiology Yes Partly No Unsure
Classification by Syndromes Yes Partly No Unsure AED Treatment & Withdrawal Details in Cohort More Information
According to a published or newly developed protocol At the discretion of treating physician Every participant on the same regimen
Randomised Not Stated/ Unclear
Quality Appraisal for All Studies
Study participation The source population or population of interest is adequately described for key characteristics Yes Partly Unsure No There is adequate participation in the study by eligible individuals Yes Partly Unsure No
Inclusion and exclusion criteria are adequately described Yes Partly Unsure No The baseline study sample (or cohort) is adequately described for key characteristics Yes Partly Unsure No
Study attrition Response rate (proportion of study sample completing the study and providing outcome data) is adequate (>80%) Yes Partly Unsure No Information on participants who dropped out of the study are described Yes Partly Unsure No
Reasons for loss to follow-up are provided and participants lost to follow-up are adequately described for key characteristics Yes Partly Unsure No There are no important differences between those who completed the study and those who did not Yes Partly Unsure No Prognostic factor measurement
The prognostic factors measured are clearly defined or described where necessary Yes Partly Unsure No The prognostic factor measure and method are adequately valid and reliable to limit misclassification bias Yes Partly Unsure No
The method and setting of measurement are the same or similar for all study participants Yes Partly Unsure No
Adequate proportion of the study sample has complete data for prognostic factors Yes Partly Unsure No Outcome measurement
A clear definition of the outcome of interest is provided, including duration of follow-up and level and extent of the outcome construct Yes Partly Unsure No The outcome measure and method used are adequately valid and reliable to limit misclassification bias Yes Partly Unsure No The method and setting of measurement are the same or similar for all study participants Yes Partly Unsure No
The outcome measurement was blinded to prognostic factor measure or vice versa Yes Partly Unsure No
145
Analysis There is sufficient presentation of data to assess the adequacy of the analysis Yes Partly Unsure No
The strategy for model building (i.e. inclusion of variables) is appropriate and is based on a conceptual framework or model Yes Partly Unsure No The selected model is adequate for the design of the study Yes Partly Unsure No There is no selective reporting of results Yes Partly Unsure No
Quality Appraisal for Models Statistical validity
The mathematical technique used to derive the model is adequately justified Yes Partly Unsure No There is no over-fitting the data with too few outcome events per predictor variable (i.e. EPV at least 10) Yes Partly Unsure No There is an account of how predictor variables were selected (i.e. univariate association, stepwise analysis, and assessment for colinearity) Yes Partly Unsure No
There is the use of appropriate imputation methods for missing data Yes Partly Unsure No Internal validation & Reproducibility of predictive variables
The method of ensuring/verifying interobserver reliability of predictor variables is described Yes Partly Unsure No The final model was validated on the data that was used to generate it to assess its performance Yes Partly Unsure No The method used for internal validation of the model is adequately described Yes Partly Unsure No
The model produced accurate predictions on the patients that were used to generate it Yes Partly Unsure No External validation & Effects of use on clinical practice
The final model was/has been prospectively validated on new cohort(s) of patients, producing accurate results Yes Partly Unsure No The prospective validation study was conducted by different clinicians from those who were used to develop the model Yes Partly Unsure No The model produced accurate predictions on the patients other than those used to generate it Yes Partly Unsure No RCTs have shown that physicians and patients are willing to use the model based on its effect on clinical outcomes Yes Partly Unsure No
Sensibility
The model is clinically sensible as no obviously clinically important items are missing Yes Partly Unsure No
The variables included in the model are easily assessable clinically Yes Partly Unsure No The model is easy and simple to use arithmetically as it does not require much time and extensive calculations Yes Partly Unsure No The model prescribes a course of action to be taken following certain probability of outcome Yes Partly Unsure No
ROC Curve Yes No
Predictive Performance – Development Data
Sensitivity - Specificity - Positive Predictive Value - Negative Predictive Value -
Correct Prediction –
Predictive Performance – External Validation
Sensitivity - Specificity - Positive Predictive Value -
Negative Predictive Value - Correct Prediction –
146
Data Extraction and Quality Appraisal Form Version 3 (Final) Data Extraction
First Author, Year, Title & Journal: ___________________________________________________________________________________ Study Design
Prospective Retrospective Mixed Unclear
Type of Study
Randomised Controlled Trial Non-Randomised Controlled Trial Cohort Study Nested Cohort Study Case Control Study
Study Setting Hospital Community
Study Recruitment Period From___________________ To________________________ (n= )
Age Bracket ________________, Mean ___________ ± _________SD
Follow up Initial Population ________, Final_________, %__________ Not Stated/ Unclear Length of Reported Follow up
Minimal_________, Mean_________, Median_________, Maximum__________ Not Stated/ Unclear Outcome More Information
Seizure Outcome Measure
Remission Time to Event
Intractability/ Refractoriness Status at specified points within follow-up timeline Others ______________________________________
Proportion of Patients with Outcome ______________________%___________ Not Stated/ Unclear
Independent Predictors of Seizure Outcome (with Regression Coefficient, Odds Ratio or Hazard Ratio and 95% CI) 1.
2. 3. 4.
5. 6.
147
Epilepsy Subtypes Included/Excluded More Information All Included - Unselected
Not stated - Not sure Selected
Method of Multivariate Analysis More Information Logistic Regression Model Cox Proportional Hazards Model
ILAE Reference Standards More Information Definition of Epilepsy Yes Partly No Unsure Classification of Seizures Yes Partly No Unsure
Classification by Aetiology Yes Partly No Unsure Classification by Syndromes Yes Partly No Unsure
AED Treatment & Withdrawal Details in Cohort More Information According to a published protocol At the discretion of treating physician
Every participant on the same regimen Randomised Not Stated/ Unclear
Reporting Characteristics of Studies Justification of sample size Yes No Report of test for colinearity Yes No
Report of test for interactions Yes No
Model assumptions tested Yes No Mention of missing data Yes No
Criteria for Variable in Model Yes No
Report on loss to follow up Yes No Statistical Package Stated Yes No
Type of stepwise analysis Yes No How continuous variables were treated None Considered Kept continuous Categorised _____________________________
Quality Appraisal for All Studies
Study participation The study sample represents the population of interest on key characteristics, sufficient to limit potential bias to the results
The source population is adequately described for key characteristics Yes Partly Unsure No Inclusion criteria and baseline cohort described Yes Partly Unsure No There is adequate participation in the study by eligible individuals Yes Partly Unsure No
Study attrition Loss to follow-up is not associated with key characteristics (i.e., the study data adequately represent the sample)
Response rate (proportion of study sample completing the study and providing outcome data) is adequate Yes Partly Unsure No Reasons for loss to follow-up are provided and those lost to follow-up are adequately described Yes Partly Unsure No There are no important differences between those who completed the study and those who did not Yes Partly Unsure No
Prognostic factor measurement The prognostic factor of interest is adequately measured in study participants
The prognostic factors measured are clearly defined Yes Partly Unsure No The measure methods adequately limit misclassification bias Yes Partly Unsure No
Adequate proportion of the study sample has complete data Yes Partly Unsure No The continuous variables were handled appropriately Yes Partly Unsure No
148
Outcome measurement
The outcome of interest is adequately measured in study participants A clear definition of the outcome is provided Yes Partly Unsure No
The measure method adequately limit misclassification bias Yes Partly Unsure No The outcome was blinded to prognostic factor or vice versa Yes Partly Unsure No Analysis
The statistical analysis is appropriate for the design of the study, limiting potential for presentation of invalid results There is sufficient presentation of data to assess the adequacy of the analysis Yes Partly Unsure No
The strategy for model building (i.e. inclusion of variables) is described Yes Partly Unsure No There is no over-fitting the data Yes Partly Unsure No
Quality Appraisal for Externally Validated Models
Validation The final model was validated on the data that was used to generate it to assess its performance Yes No The final model was/has been prospectively validated on new cohort(s) of patients, producing accurate results Yes No
RCTs have shown that physicians and patients are willing to use the model based on its effect on clinical outcomes Yes No Sensibility The model is clinically sensible (no important items are missing ) Yes No
The variables included in the model are easily assessable clinically Yes No The model is easy and simple to use arithmetically Yes No The model prescribes a course of action to be taken Yes No
ROC Curve Characteristics Presented Yes No
Predictive Performance – Development Data Sensitivity -
Specificity - Positive Predictive Value - Negative Predictive Value -
Correct Prediction – Predictive Performance – External Validation Sensitivity -
Specificity - Positive Predictive Value -
Negative Predictive Value -
Correct Prediction -
149
7.4 Appendix IV: ITEMS FOR THE ASSESSSMENT OF BIAS IN PROGNOSIS STUDIES*
* Hayden JA, Cote P, Bombardier C. Evaluation of the quality of prognosis studies in systematic reviews. Annals of Internal Medicine2006 Mar 21;144(6):427-37.
150
7.5 Appendix V: ITEMS TO BE CONSIDERED IN REPORTING OBSERVATIONAL STUDIES*
* von Elm E, Altman DG, Egger M, Pocock SJ, GÃtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. Annals of Internal Medicine2007 October 16, 2007; 147(8):573-7.
151
152
8 REFERENCES
1. Anonymous. Guidelines for epidemiologic studies on epilepsy. Commission on Epidemiology and Prognosis, International League Against Epilepsy. Epilepsia1993 Jul-Aug;34(4):592-6.
2. Anonymous. ILAE Commission Report. The epidemiology of the epilepsies: future directions. International League Against Epilepsy. Epilepsia1997 May;38(5):614-8.
3. WHO. Atlas: Epilepsy Care in the World. Geneva: World Health Organization; 2005.
4. Leonardi M, Ustun TB. The Global Burden of Epilepsy. Epilepsia2002;43(s6):21-5.
5. Bromfield E, Cavazos J, Sirven J, editors. An Introduction to Epilepsy. West Hartford (CT): American Epilepsy Society; 2006.
6. Hauser WA, Annegers JF, Kurland LT. Incidence of epilepsy and unprovoked seizures in Rochester, Minnesota: 1935-1984. Epilepsia1993 May-Jun;34(3):453-68.
7. Tallis R, Hall G, Craig I, Dean A. How common are epileptic seizures in old age? Age & Ageing1991 Nov;20(6):442-8.
8. Shorvon S. Epilepsy. Oxford: Oxford University Press; 2009.
9. Berg AT, Berkovic SF, Brodie MJ, Buchhalter J, Cross JH, Boas WvE, et al. Revised terminology and concepts for organization of seizures and epilepsies: Report of the ILAE Commission on Classification and Terminology, 2005-2009. Epilepsia2010;51(4):676-85.
10. Dekker P. Epilepsy: A Manual for Medical and Clinical Officers in Africa. Geneva: World Health Organization; 2002. Available from: http://www.who.int/mental_health/media/en/639.pdf.
11. Anonymous. Proposal for revised classification of epilepsies and epileptic syndromes. Commission on Classification and Terminology of the International League Against Epilepsy. Epilepsia1989 Jul-Aug;30(4):389-99.
12. Avanzini G. Of Cabbages and Kings: Do We Really Need a Systematic Classification of Epilepsies? Epilepsia2003;44:1-13.
13. Berg AT, Blackstone NW. Of Cabbages and Kings: Perspectives on Classification from the Field of Systematics. Epilepsia2003;44:1-13.
14. Engel J. Reply to "Of Cabbages and Kings: Some Considerations on Classifications, Diagnostic Schemes, Semiology, and Concepts". Epilepsia2003;44:1-13.
15. Fisher RS. Editors Introduction: Cabbages and Kings in the Classification of Seizures and the Epilepsies. Epilepsia2003;44:1-13.
153
16. Lüders H, Najm I, Wyllie E. Reply to "Of Cabbages and Kings: Some Considerations on Classifications, Diagnostic Schemes, Semiology, and Concepts". Epilepsia2003;44:1-13.
17. Wolf P. Of Cabbages and Kings: Some Considerations on Classifications, Diagnostic Schemes, Semiology, and Concepts. Epilepsia2003;44:1-13.
18. Anonymous. Proposal for revised clinical and electroencephalographic classification of epileptic seizures. From the Commission on Classification and Terminology of the International League Against Epilepsy. Epilepsia1981 Aug;22(4):489-501.
19. Sridharan R. Epidemiology of epilepsy Current Science 2002;82(6).
20. Sander JW, Shorvon SD. Incidence and prevalence studies in epilepsy and their methodological problems: a review. Journal of Neurology, Neurosurgery & Psychiatry1987 Jul;50(7):829-39.
21. Berg AT, Shinnar S, Levy SR, Testa FM, Smith-Rapaport S, Beckerman B, et al. Two-year remission and subsequent relapse in children with newly diagnosed epilepsy. Epilepsia2001;42(12):1553-62.
22. Delgado-RodrÃguez M, Llorca J. Bias. Journal of Epidemiology and Community Health2004 August 2004;58(8):635-41.
23. Moons KG, Royston P, Vergouwe Y, Grobbee DE, Altman DG, Moons KGM, et al. Prognosis and prognostic research: what, why, and how? BMJ2009;338:b375.
24. Shinnar S, Berg AT, O'Dell C, Newstein D, Moshe SL, Hauser WA. Predictors of multiple seizures in a cohort of children prospectively followed from the time of their first unprovoked seizure. Annals of Neurology2000 Aug;48(2):140-7.
25. Sillanpaa M, Schmidt D, Sillanpaa M, Schmidt D. Natural history of treated childhood-onset epilepsy: prospective, long-term population-based study. Brain2006 Mar;129(Pt 3):617-24.
26. Iezzoni LI. Statistically derived predictive models. Caveat emptor. Journal of General Internal Medicine1999 Jun;14(6):388-9.
27. Altman DG, Vergouwe Y, Royston P, Moons KG, Altman DG, Vergouwe Y, et al. Prognosis and prognostic research: validating a prognostic model. BMJ2009;338:b605.
28. Rochon PA, Gurwitz JH, Sykora K, Mamdani M, Streiner DL, Garfinkel S, et al. Reader's guide to critical appraisal of cohort studies: 1. Role and design. BMJ2005 April 16, 2005;330(7496):895-7.
29. Velissaris SL, Wilson SJ, Saling MM, Newton MR, Berkovic SF, Velissaris SL, et al. The psychological impact of a newly diagnosed seizure: losing and restoring perceived control. Epilepsy & Behavior2007 Mar;10(2):223-33.
154
30. Jacoby A, Baker GA, Steen N, Potts P, Chadwick DW. The clinical course of epilepsy and its psychosocial correlates: findings from a U.K. Community study. Epilepsia1996 Feb;37(2):148-61.
31. Baker GA, Jacoby A, Buck D, Stalgis C, Monnet D. Quality of life of people with epilepsy: a European study. Epilepsia1997 Mar;38(3):353-62.
32. Sillanpaa M, Jalava M, Kaleva O, Shinnar S. Long-term prognosis of seizures with onset in childhood. New England Journal of Medicine1998 Jun 11;338(24):1715-22.
33. Nilsson L, Ahlbom A, Farahmand BY, Asberg M, Tomson T, Nilsson L, et al. Risk factors for suicide in epilepsy: a case control study. Epilepsia2002 Jun;43(6):644-51.
34. DeLorenzo RJ, Hauser WA, Towne AR, Boggs JG, Pellock JM, Penberthy L, et al. A prospective, population-based epidemiologic study of status epilepticus in Richmond, Virginia. Neurology1996 Apr;46(4):1029-35.
35. Tomson T, Nashef L, Ryvlin P. Sudden unexpected death in epilepsy: current knowledge and future directions. The Lancet Neurology2008;7(11):1021-31.
36. Opeskin K, Berkovic SF, Opeskin K, Berkovic SF. Risk factors for sudden unexpected death in epilepsy: a controlled prospective study based on coroners cases. Seizure2003 Oct;12(7):456-64.
37. Opeskin K, Harvey AS, Cordner SM, Berkovic SF. Sudden unexpected death in epilepsy in Victoria. Journal of Clinical Neuroscience2000 Jan;7(1):34-7.
38. Docherty M, Smith R. The case for structuring the discussion of scientific papers. BMJ1999 May 8, 1999;318(7193):1224-5.
39. Hayden JA, Cote P, Bombardier C, Hayden JA, Cote P, Bombardier C. Evaluation of the quality of prognosis studies in systematic reviews. Annals of Internal Medicine2006 Mar 21;144(6):427-37.
40. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. Annals of Internal Medicine2007 October 16, 2007;147(8):573-7.
41. Fletcher R, Fletcher S, Wagner E. Clinical epidemiology—the essentials. Baltimore/London: Williams & Wilkins; 1988.
42. Hemingway H, Riley R, Altman D. Ten steps towards improving prognosis research. BMJ2009;339:b4184
43. Windeler J. Prognosis - what does the clinician associate with this notion? Stat Med2000 Feb 29;19(4):425-30.
155
44. Berg AT, Shinnar S. The risk of seizure recurrence following a first unprovoked seizure: a quantitative review. Neurology1991 Jul;41(7):965-72.
45. Altman D, Lyman G. Methodological challenges in the evaluation of prognostic factors in breast cancer. Breast Cancer Research and Treatment1998;52(1):289-303.
46. Plato. Book XI. In: B. Jowett, Editor, Laws. New York: Charles Scribner; 1871.
47. Magiorkinis E, Sidiropoulou K, Diamantis A, Magiorkinis E, Sidiropoulou K, Diamantis A. Hallmarks in the history of epilepsy: epilepsy in antiquity. Epilepsy Behav2010 Jan;17(1):103-8.
48. Hippocrate. De la maladie sacrée, livre 6. In: E. Littré, Editor, Oeuvres complètes d’ Hippocrate. Paris Baillière; 1849.
49. Gowers W. Epilepsy and other chronic convulsive disorders London: Churchill Livingstone; 1881.
50. Rodin EA, Rhodes RJ, Velarde NN. The prognosis for patients with epilepsy. J Occup Med1965 Nov;7(11):560-3.
51. Kwan P, Sander JW. The natural history of epilepsy: an epidemiological view. Journal of Neurology, Neurosurgery & Psychiatry2004 Oct;75(10):1376-81.
52. Meyer A-C, Dua T, Ma J, Saxena S, Birbeck G. Global disparities in the epilepsy treatment gap: a systematic review. Bulletin of the World Health Organization 2010;88(4):260-6.
53. Watts AE. The natural history of untreated epilepsy in a rural community in Africa. Epilepsia1992 May-Jun;33(3):464-8.
54. Placencia M, Shorvon SD, Paredes V, Bimos C, Sander JW, Suarez J, et al. Epileptic seizures in an Andean region of Ecuador. Incidence and prevalence and regional variation. Brain1992 Jun;115(Pt 3):771-82.
55. Wang WZ, Wu JZ, Wang DS, Dai XY, Yang B, Wang TP, et al. The prevalence and treatment gap in epilepsy in China: an ILAE/IBE/WHO study. Neurology2003 May 13;60(9):1544-5.
56. Nicoletti A, Sofia V, Vitale G, Bonelli SI, Bejarano V, Bartalesi F, et al. Natural history and mortality of chronic epilepsy in an untreated population of rural Bolivia: a follow-up after 10 years. Epilepsia2009 Oct;50(10):2199-206.
57. WHO. [11 May 2010]; Available from: http://www.who.int/countries/chn/en/.
58. Keranen T, Riekkinen PJ. Remission of seizures in untreated epilepsy. BMJ1993 Aug 21;307(6902):483.
59. Zielihski JJ. Epileptics Not in Treatment*. Epilepsia1974;15(2):203-10.
156
60. Tuan NA, Tomson T, Allebeck P, Chuc NT, Cuong le Q, Tuan NA, et al. The treatment gap of epilepsy in a rural district of Vietnam: a study from the EPIBAVI project.[Erratum appears in Epilepsia. 2009 Nov;50(11):2506]. Epilepsia2009 Oct;50(10):2320-3.
61. Loiseau PJ. Clinical experience with new antiepileptic drugs: antiepileptic drugs in Europe. Epilepsia1999;40 Suppl 6:S3-8; discussion S73-4.
62. Perucca E. The new generation of antiepileptic drugs: advantages and disadvantages. British Journal of Clinical Pharmacology1996 Nov;42(5):531-43.
63. Abduljabbar M, Ogunniyi A, Daif AK, Al-Tahan A, Al-Bunyan M, Al-Rajeh S. Epilepsy classification and factors associated with control in Saudi adult patients. Seizure1998 Dec;7(6):501-4.
64. Collaborative Group for the Study of Epilepsy. Prognosis of epilepsy in newly referred patients: a multicenter prospective study of the effects of monotherapy on the long-term course of epilepsy. Collaborative Group for the Study of Epilepsy. Epilepsia1992 Jan-Feb;33(1):45-51.
65. Arts WFM, Geerts AT, Brouwer OF, Peters ACB, Stroink H, Van Donselaar CA. The early prognosis of epilepsy in childhood: The prediction of a poor outcome. The Dutch Study of Epilepsy in Childhood. Epilepsia1999;40(6):726-34.
66. Arts WFM, Brouwer OF, Peters ACB, Stroink H, Peeters EAJ, Schmitz PIM, et al. Course and prognosis of childhood epilepsy: 5-Year follow-up of the Dutch study of epilepsy in childhood. Brain2004;127(8):1774-84.
67. Casetta I, Granieri E, Monetti VC, Tola MR, Paolino E, Malagu S, et al. Prognosis of childhood epilepsy: a community-based study in Copparo, Italy. Neuroepidemiology1997;16(1):22-8.
68. Lossius MI, Stavem K, Gjerstad L. Predictors for recurrence of epileptic seizures in a general epilepsy population. Seizure1999 Dec;8(8):476-9.
69. Sillanpaa M. Children with epilepsy as adults: outcome after 30 years of follow-up. Acta Paediatr Scand Suppl1990;368:1-78.
70. Sillanpaa M, Schmidt D, Sillanpaa M, Schmidt D. Early seizure frequency and aetiology predict long-term medical outcome in childhood-onset epilepsy. Brain2009 Apr;132(Pt 4):989-98.
71. Camfield C, Camfield P, Gordon K, Smith B, Dooley J. Outcome of childhood epilepsy: a population-based study with a simple predictive scoring system for those treated with medication. Journal of Pediatrics1993 Jun;122(6):861-8.
72. Oskoui M, Webster RI, Zhang X, Shevell MI, Oskoui M, Webster RI, et al. Factors predictive of outcome in childhood epilepsy. J Child Neurol2005 Nov;20(11):898-904.
157
73. Berg AT, Shinnar S, Levy SR, Testa FM, Smith-Rapaport S, Beckerman B. Early development of intractable epilepsy in children: A prospective study. Neurology2001;56(11):1445-52.
74. Ramos-Lizana J, Aguilera-Lopez P, Aguirre-Rodriguez J, Cassinello-Garcia E. Early prediction of refractory epilepsy in childhood. Seizure2009;18(6):412-6.
75. Casetta I, Granieri E, Monetti VC, Gilli G, Tola MR, Paolino E, et al. Early predictors of intractability in childhood epilepsy: a community-based case-control study in Copparo, Italy. Acta Neurol Scand1999 Jun;99(6):329-33.
76. Kwong KL, Sung WY, Wong SN, So KT, Kwong KL, Sung WY, et al. Early predictors of medical intractability in childhood epilepsy. Pediatr Neurol2003 Jul;29(1):46-52.
77. Sillanpaa M. Remission of seizures and predictors of intractability in long-term follow-up. Epilepsia1993;34(5):930-6.
78. Berg AT, Levy SR, Novotny EJ, Shinnar S. Predictors of intractable epilepsy in childhood: A case-control study. Epilepsia1996;37(1):24-30.
79. Ko TS, Holmes GL. EEG and clinical predictors of medically intractable childhood epilepsy. Clin Neurophysiol1999 Jul;110(7):1245-51.
80. Camfield P, Camfield C, Camfield P, Camfield C. Childhood epilepsy: what is the evidence for what we think and what we do? J Child Neurol2003 Apr;18(4):272-87.
81. Kwan P, Brodie MJ. Early identification of refractory epilepsy. New England Journal of Medicine2000 Feb 3;342(5):314-9.
82. Hakkarainen H. Carbamazepine and diphenylhydantoin as monotherapy or in combination in the treatment of adult epilepsy. Neurology1980;48:1010-4.
83. Tanganelli P, Regesta G. Vigabatrin vs. carbamazepine monotherapy in newly diagnosed focal epilepsy: a randomized response conditional cross-over study. Epilepsy Research1996 Nov;25(3):257-62.
84. Wiebe S, Wiebe S. Early epilepsy surgery. Current Neurology & Neuroscience Reports2004 Jul;4(4):315-20.
85. Camfield PR, Camfield CS, Smith EC, Tibbles JA. Newly treated childhood epilepsy: a prospective study of recurrences and side effects. Neurology1985 May;35(5):722-5.
86. de Silva M, MacArdle B, McGowan M, Hughes E, Stewart J, Neville BG, et al. Randomised comparative monotherapy trial of phenobarbitone, phenytoin, carbamazepine, or sodium valproate for newly diagnosed childhood epilepsy. Lancet1996 Mar 16;347(9003):709-13.
158
87. Ramsay RE, Wilder BJ, Berger JR, Bruni J. A double-blind study comparing carbamazepine with phenytoin as initial seizure therapy in adults. Neurology1983 Jul;33(7):904-10.
88. Glauser T, Ben-Menachem E, Bourgeois B, Cnaan A, Chadwick D, Guerreiro C, et al. ILAE treatment guidelines: evidence-based analysis of antiepileptic drug efficacy and effectiveness as initial monotherapy for epileptic seizures and syndromes. Epilepsia2006 Jul;47(7):1091-3.
89. Schmidt D. Reduction of two-drug therapy in intractable epilepsy. Epilepsia1983 Jun;24(3):368-76.
90. Camfield C, Camfield P, Gordon K, Dooley J. A Population-based Study of Childhood Epilepsy That Does Not Recur on Treatment and Remits When Medication Is Stopped: Smooth Sailing Epilepsy. Ann Neurol1990 Sept;28(3):471-2.
91. Huttenlocher PR, Hapke RJ. A follow-up study of intractable seizures in childhood. Ann Neurol1990 Nov;28(5):699-705.
92. Altunbasak S, Incecik F, Herguner O, Refik Burgut H, Altunbasak S, Incecik F, et al. Prognosis of patients with seizures occurring in the first 2 years. J Child Neurol2007 Mar;22(3):307-13.
93. Banu SH, Khan NZ, Hossain M, Jahan A, Parveen M, Rahman N, et al. Profile of childhood epilepsy in Bangladesh. Developmental Medicine & Child Neurology2003;45(7):477-82.
94. Chawla S, Aneja S, Kashyap R, Mallika V, Chawla S, Aneja S, et al. Etiology and clinical predictors of intractable epilepsy. Pediatric Neurology2002 Sep;27(3):186-91.
95. Collaborative Group for the Study of Epilepsy. Prognosis of epilepsy in newly referred patients: a multicenter prospective study. Collaborative Group for the Study of Epilepsy. Epilepsia1988 May-Jun;29(3):236-43.
96. Del Felice A, Beghi E, Boero G, La Neve A, Bogliun G, De Palo A, et al. Early versus late remission in a cohort of patients with newly diagnosed epilepsy. Epilepsia2010;51(1):37-42.
97. Hui ACF, Wong A, Wong HC, Man BL, Au-Yeung KM, Wong KS. Refractory epilepsy in a Chinese population. Clinical Neurology and Neurosurgery2007;109(8):672-5.
98. Kim LG, Johnson TL, Marson AG, Chadwick DW, group MMS, Kim LG, et al. Prediction of risk of seizure recurrence after a single seizure and early epilepsy: further results from the MESS trial.[Erratum appears in Lancet Neurol. 2006 May;5(5):383]. Lancet Neurology2006 Apr;5(4):317-22.
99. Lindsten H, Stenlund H, Forsgren L. Seizure recurrence in adults after a newly diagnosed unprovoked epileptic seizure. Acta Neurol Scand2001 Oct;104(4):202-7.
159
100. Lohani S, Devkota UP, Rajbhandari H. Predictors of unfavourable seizure outcome in patients with epilepsy in Nepal. Canadian Journal of Neurological Sciences2010;37(1):76-80.
101. MacDonald BK, Johnson AL, Goodridge DM, Cockerell OC, Sander JW, Shorvon SD. Factors predicting prognosis of epilepsy after presentation with seizures.[Erratum appears in Ann Neurol 2001 Apr;49(4):547].[Erratum appears in Ann Neurol 2001 Dec;50(6):830]. Annals of Neurology2000 Dec;48(6):833-41.
102. Rantala H, Ingalsuo H. Occurrence and outcome of epilepsy in children younger than 2 years. Journal of Pediatrics1999;135(6):761-4.
103. Semah F, Picot MC, Adam C, Broglin D, Arzimanoglou A, Bazin B, et al. Is the underlying cause of epilepsy a major prognostic factor for recurrence? Neurology1998 Nov;51(5):1256-62.
104. Shafer SQ, Hauser WA, Annegers JF, Klass DW. EEG and other early predictors of epilepsy remission: a community study. Epilepsia1988;29(5):590-600.
105. Sillanpaa M, Camfield P, Camfield C. Predicting long-term outcome of childhood epilepsy in Nova Scotia, Canada, and Turku, Finland. Validation of a simple scoring system. Archives of Neurology1995 Jun;52(6):589-92.
106. Geelhoed M, Boerrigter AO, Camfield P, Geerts AT, Arts W, Smith B, et al. The accuracy of outcome prediction models for childhood-onset epilepsy. Epilepsia2005 Sep;46(9):1526-32.
107. Geerts AT, Arts WF, Brouwer OF, Peters AC, Peeters EA, Stroink H, et al. Validation of two prognostic models predicting outcome at two years after diagnosis in a new cohort of children with epilepsy: the Dutch Study of Epilepsy in Childhood. Epilepsia2006 Jun;47(6):960-5.
108. Temkin NR, Dikmen SS, Anderson GD, Wilensky AJ, Holmes MD, Cohen W, et al. Valproate therapy for prevention of posttraumatic seizures: a randomized trial. Journal of Neurosurgery1999 Oct;91(4):593-600.
109. Temkin NR, Dikmen SS, Wilensky AJ, Keihm J, Chabal S, Winn HR. A randomized, double-blind study of phenytoin for the prevention of post-traumatic seizures. New England Journal of Medicine1990 Aug 23;323(8):497-502.
110. Foy PM, Chadwick DW, Rajgopalan N, Johnson AL, Shaw MD. Do prophylactic anticonvulsant drugs alter the pattern of seizures after craniotomy? Journal of Neurology, Neurosurgery & Psychiatry1992 Sep;55(9):753-7.
111. Musicco M, Beghi E, Solari A, Viani F. Treatment of first tonic-clonic seizure does not improve the prognosis of epilepsy. First Seizure Trial Group (FIRST Group). Neurology1997 Oct;49(4):991-8.
160
112. Lhatoo SD, Sander JW. Stopping drug therapy in epilepsy. Current Pharmaceutical Design2000 May;6(8):861-3.
113. Britton JW, Britton JW. Antiepileptic drug withdrawal: literature review. Mayo Clinic Proceedings2002 Dec;77(12):1378-88.
114. Berg AT, Shinnar S. Relapse following discontinuation of antiepileptic drugs: a meta-analysis. Neurology1994 Apr;44(4):601-8.
115. Anonymous. Randomised study of antiepileptic drug withdrawal in patients in remission. Medical Research Council Antiepileptic Drug Withdrawal Study Group. Lancet1991 May 18;337(8751):1175-80.
116. Chadwick D. Does withdrawal of different antiepileptic drugs have different effects on seizure recurrence? Further results from the MRC Antiepileptic Drug Withdrawal Study. Brain1999 Mar;122(Pt 3):441-8.
117. Chadwick D, Taylor J, Johnson T. Outcomes after seizure recurrence in people with well-controlled epilepsy and the factors that influence it. The MRC Antiepileptic Drug Withdrawal Group. Epilepsia1996 Nov;37(11):1043-50.
118. Stephen LJ, Kwan P, Brodie MJ. Does the cause of localisation-related epilepsy influence the response to antiepileptic drug treatment? Epilepsia2001 Mar;42(3):357-62.
119. Hauser WA, Lee JR, Hauser WA, Lee JR. Do seizures beget seizures? Progress in Brain Research2002;135:215-9.
120. Camfield C, Camfield P, Gordon K, Dooley J. Does the number of seizures before treatment influence ease of control or remission of childhood epilepsy? Not if the number is 10 or less. Neurology1996 Jan;46(1):41-4.
121. Shorvon SD, Reynolds EH. Early prognosis of epilepsy. Br Med J (Clin Res Ed)1982 Dec 11;285(6356):1699-701.
122. Berg AT, Shinnar S. Do seizures beget seizures? An assessment of the clinical evidence in humans. J Clin Neurophysiol1997 Mar;14(2):102-10.
123. Multani P, Myers RH, Blume HW, Schomer DL, Sotrel A. Neocortical Dendritic Pathology in Human Partial Epilepsy: A Quantitative Golgi Study. Epilepsia1994;35(4):728-36.
124. Liu RS, Lemieux L, Bell GS, Hammers A, Sisodiya SM, Bartlett PA, et al. Progressive neocortical damage in epilepsy. Ann Neurol2003 Mar;53(3):312-24.
125. Fuerst D, Shah J, Shah A, Watson C, Fuerst D, Shah J, et al. Hippocampal sclerosis is a progressive disorder: a longitudinal volumetric MRI study. Annals of Neurology2003 Mar;53(3):413-6.
161
126. Briellmann RS, Berkovic SF, Syngeniotis A, King MA, Jackson GD, Briellmann RS, et al. Seizure-associated hippocampal volume loss: a longitudinal magnetic resonance study of temporal lobe epilepsy. Annals of Neurology2002 May;51(5):641-4.
127. Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatrica Scandinavica1983 Jun;67(6):361-70.
128. Thompson AW, Miller JW, Katon W, Chaytor N, Ciechanowski P. Sociodemographic and clinical factors associated with depression in epilepsy. Epilepsy Behav2009;14(4):655-60.
129. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. Journal of General Internal Medicine2001 Sep;16(9):606-13.
130. Camfield C, Camfield P, Smith B, Gordon K, Dooley J. Biologic factors as predictors of social outcome of epilepsy in intellectually normal children: a population-based study. Journal of Pediatrics1993 Jun;122(6):869-73.
131. Jokeit H, Ebner A. Long term effects of refractory temporal lobe epilepsy on cognitive abilities: a cross sectional study. Journal of Neurology, Neurosurgery & Psychiatry1999 Jul;67(1):44-50.
132. Nolan MA, Redoblado MA, Lah S, Sabaz M, Lawson JA, Cunningham AM, et al. Intelligence in childhood epilepsy syndromes. Epilepsy Research2003;53(1-2):139-50.
133. Pavone P, Bianchini R, Trifiletti RR, Incorpora G, Pavone A, Parano E. Neuropsychological assessment in children with absence epilepsy. Neurology2001 Apr 24;56(8):1047-51.
134. Meador KJ, Loring DW, Moore EE, Thompson WO, Nichols ME, Oberzan RE, et al. Comparative cognitive effects of phenobarbital, phenytoin, and valproate in healthy adults. Neurology1995 Aug;45(8):1494-9.
135. Trimble MR, Thompson PJ. Anticonvulsant drugs, cognitive function, and behavior. Epilepsia1983;24 Suppl 1:S55-63.
136. Meador KJ, Meador KJ. Cognitive outcomes and predictive factors in epilepsy. Neurology2002 Apr 23;58(8 Suppl 5):S21-6.
137. Kwan P, Brodie MJ. Neuropsychological effects of epilepsy and antiepileptic drugs. Lancet2001 Jan 20;357(9251):216-22.
138. Gilliam FG, Fessler AJ, Baker G, Vahle V, Carter J, Attarian H. Systematic screening allows reduction of adverse antiepileptic drug effects: a randomized trial. Neurology2004 Jan 13;62(1):23-7.
139. Dlugos DJ. The early identification of candidates for epilepsy surgery. Arch Neurol2001 Oct;58(10):1543-6.
140. Heaney DC, Begley CE, Heaney DC, Begley CE. Economic evaluation of epilepsy treatment: a review of the literature. Epilepsia2002;43 Suppl 4:10-6.
162
141. Wiebe S, Gafni A, Blume WT, Girvin JP. An economic evaluation of surgery for temporal lobe epilepsy. Journal of Epilepsy1995;8(3):227-35.
142. King JT, Jr., Sperling MR, Justice AC, O'Connor MJ. A cost-effectiveness analysis of anterior temporal lobectomy for intractable temporal lobe epilepsy. Journal of Neurosurgery1997 Jul;87(1):20-8.
143. Langfitt JT. Cost-effectiveness of anterotemporal lobectomy in medically intractable complex partial epilepsy. Epilepsia1997 Feb;38(2):154-63.
144. Salanova V, Markand O, Worth R. Temporal lobe epilepsy surgery: outcome, complications, and late mortality rate in 215 patients. Epilepsia2002 Feb;43(2):170-4.
145. Nilsson L, Ahlbom A, Farahmand BY, Tomson T, Nilsson L, Ahlbom A, et al. Mortality in a population-based cohort of epilepsy surgery patients. Epilepsia2003 Apr;44(4):575-81.
146. Meyer FB, Marsh WR, Laws ER, Jr., Sharbrough FW. Temporal lobectomy in children with epilepsy. Journal of Neurosurgery1986 Mar;64(3):371-6.
147. Helmstaedter C, Kurthen M, Lux S, Reuber M, Elger CE, Helmstaedter C, et al. Chronic epilepsy and cognition: a longitudinal study in temporal lobe epilepsy. Annals of Neurology2003 Oct;54(4):425-32.
148. Helmstaedter C, Reuber M, Elger CC, Helmstaedter C, Reuber M, Elger CCE. Interaction of cognitive aging and memory deficits related to epilepsy surgery. Annals of Neurology2002 Jul;52(1):89-94.
149. Westerveld M, Sass KJ, Chelune GJ, Hermann BP, Barr WB, Loring DW, et al. Temporal lobectomy in children: cognitive outcome. Journal of Neurosurgery2000 Jan;92(1):24-30.
150. Davies KG, Bell BD, Bush AJ, Wyler AR. Prediction of verbal memory loss in individuals after anterior temporal lobectomy. Epilepsia1998 Aug;39(8):820-8.
151. Engel J, Jr., Wiebe S, French J, Sperling M, Williamson P, Spencer D, et al. Practice parameter: temporal lobe and localized neocortical resections for epilepsy: report of the Quality Standards Subcommittee of the American Academy of Neurology, in association with the American Epilepsy Society and the American Association of Neurological Surgeons. Neurology2003 Feb 25;60(4):538-47.
152. Dlugos DJ, Sammel MD, Strom BL, Farrar JT. Response to first drug trial predicts outcome in childhood temporal lobe epilepsy. Neurology2001 Dec 26;57(12):2259-64.
153. Dlugos DJ, Buono RJ, Dlugos DJ, Buono RJ. Predicting outcome of initial treatment with carbamazepine in childhood focal epilepsy. Pediatr Neurol2004 May;30(5):311-5.
154. Sperling MR, O'Connor MJ, Saykin AJ, Plummer C. Temporal lobectomy for refractory epilepsy. JAMA1996 Aug 14;276(6):470-5.
163
155. Wyllie E, Comair YG, Kotagal P, Bulacio J, Bingaman W, Ruggieri P. Seizure outcome after epilepsy surgery in children and adolescents. Ann Neurol1998 Nov;44(5):740-8.
156. Engel J, Van Ness P, Rasmussen T, Ojemann L. Outcome with respect to epileptic seizures. In: Engel J, ed. Surgical Treatment of the Epilepsies. 2nd ed. New York, NY: Raven Press; 1993.
157. Engel J, Jr. Surgery for seizures. New England Journal of Medicine1996 Mar 7;334(10):647-52.
158. Fish DR, Smith SJ, Quesney LF, Andermann F, Rasmussen T. Surgical treatment of children with medically intractable frontal or temporal lobe epilepsy: results and highlights of 40 years' experience. Epilepsia1993 Mar-Apr;34(2):244-7.
159. Berkovic SF, McIntosh AM, Kalnins RM, Jackson GD, Fabinyi GC, Brazenor GA, et al. Preoperative MRI predicts outcome of temporal lobectomy: an actuarial analysis. Neurology1995 Jul;45(7):1358-63.
160. Elwes RD, Dunn G, Binnie CD, Polkey CE. Outcome following resective surgery for temporal lobe epilepsy: a prospective follow up study of 102 consecutive cases. Journal of Neurology, Neurosurgery & Psychiatry1991 Nov;54(11):949-52.
161. Lindsay J, Ounsted C, Richards P. Long-term outcome in children with temporal lobe seizures. IV: Genetic factors, febrile convulsions and the remission of seizures. Dev Med Child Neurol1980 Aug;22(4):429-39.
162. Lindsay J, Ounsted C, Richards P. Long-term outcome in children with temporal lobe seizures. V: Indications and contra-indications for neurosurgery. Dev Med Child Neurol1984 Feb;26(1):25-32.
163. Tellez-Zenteno JF, Ronquillo LH, Moien-Afshari F, Wiebe S. Surgical outcomes in lesional and non-lesional epilepsy: A systematic review and meta-analysis. Epilepsy Research2010;89(2-3):310-8.
164. Tonini C, Beghi E, Berg AT, Bogliun G, Giordano L, Newton RW, et al. Predictors of epilepsy surgery outcome: a meta-analysis. Epilepsy Research2004;62(1):75-87.
165. Engel JJ, Shewmon D. Overview. Who should be considered a surgical candidate? In: Engel J, ed. Surgical Treatment of the Epilepsies. 2nd ed. New York: Raven Pres; 1993.
166. Engel J, Jr. Etiology as a risk factor for medically refractory epilepsy: a case for early surgical intervention. Neurology1998 Nov;51(5):1243-4.
167. Wiebe S, Blume WT, Girvin JP, Eliasziw M, Effectiveness, Efficiency of Surgery for Temporal Lobe Epilepsy Study G. A randomized, controlled trial of surgery for temporal-lobe epilepsy. New England Journal of Medicine2001 Aug 2;345(5):311-8.
168. Engel J, Jr., Engel J, Jr. Surgical treatment for epilepsy: too little, too late? JAMA2008 Dec 3;300(21):2548-50.
164
169. Saneto RP, Sotero de Menezes MA, Ojemann JG, Bournival BD, Murphy PJ, Cook WB, et al. Vagus nerve stimulation for intractable seizures in children. Pediatric Neurology2006 Nov;35(5):323-6.
170. Morris GL, 3rd, Mueller WM. Long-term treatment with vagus nerve stimulation in patients with refractory epilepsy. The Vagus Nerve Stimulation Study Group E01-E05.[Erratum appears in Neurology 2000 Apr 25;54(8):1712]. Neurology1999 Nov 10;53(8):1731-5.
171. Freeman JM, Kossoff EH, Hartman AL. The Ketogenic Diet: One Decade Later. Pediatrics2007 March 1, 2007;119(3):535-43.
172. Keene DL, Keene DL. A systematic review of the use of the ketogenic diet in childhood epilepsy. Pediatric Neurology2006 Jul;35(1):1-5.
173. Henderson CB, Filloux FM, Alder SC, Lyon JL, Caplin DA, Henderson CB, et al. Efficacy of the ketogenic diet as a treatment option for epilepsy: meta-analysis. J Child Neurol2006 Mar;21(3):193-8.
174. Neal EG, Chaffe H, Schwartz RH, Lawson MS, Edwards N, Fitzsimmons G, et al. The ketogenic diet for the treatment of childhood epilepsy: a randomised controlled trial. Lancet Neurology2008 Jun;7(6):500-6.
175. Hauser WA, Rich SS, Lee JR, Annegers JF, Anderson VE. Risk of recurrent seizures after two unprovoked seizures. New England Journal of Medicine1998 Feb 12;338(7):429-34.
176. Chalmers I. Foreword In: Egger M, Smith GD, Altman D (eds) Systematic Reviews in Health Care: Meta-Analysis in Context. 2nd ed. London: BMJ Publishing Group; 2001.
177. Kleijnen J, Antes G. Book Review - Systematic Reviews in Health Care. Meta-analysis in Context.: M Egger, G Davey Smith, Doug Altman (eds). London: BMJ Books, 2001. Int J Epidemiol2002 June 1, 2002;31(3):697-.
178. Cook DJ, Sackett DL, Spitzer WO. Methodologic guidelines for systematic reviews of randomized control trials in health care from the Potsdam Consultation on Meta-Analysis. Journal of Clinical Epidemiology1995 Jan;48(1):167-71.
179. Chalmers I, Hedges LV, Cooper H. A Brief History of Research Synthesis. Eval Health Prof2002 March 1, 2002;25(1):12-37.
180. Antman EM, Lau J, Kupelnick B, Mosteller F, Chalmers TC. A Comparison of Results of Meta-analyses of Randomized Control Trials and Recommendations of Clinical Experts: Treatments for Myocardial Infarction. JAMA1992 July 8, 1992;268(2):240-8.
181. McAlister FAMDM, Clark HDMD, van Walraven CMDM, Straus SEMD, Lawson FMEMB, Moher DM, et al. The Medical Review Article Revisited: Has the Science Improved? [Miscellaneous Article]. Annals of Internal Medicine1999;131(12):947-51.
165
182. Mulrow CD. The medical review article: state of the science. Annals of Internal Medicine1987 Mar;106(3):485-8.
183. Schmidt LM, Gotzsche PC, Schmidt LM, Gotzsche PC. Of mites and men: reference bias in narrative review articles: a systematic review. Journal of Family Practice2005 Apr;54(4):334-8.
184. Ladhani S, Williams HC. The management of established postherpetic neuralgia: a comparison of the quality and content of traditional vs. systematic reviews. British Journal of Dermatology1998 Jul;139(1):66-72.
185. Dickersin K. Systematic reviews in epidemiology: why are we so far behind? Int J Epidemiol2002 February 1, 2002;31(1):6-12.
186. Peters CC. Summary of the Penn State Experiments on the Influence of Instruction in Character Education. Journal of Educational Sociology1933;7(4):269-72.
187. Glass GV. Primary, Secondary, and Meta-Analysis of Research. Educational Researcher1976;5(10):3-8.
188. Chalmers I, Trohler U. Helping physicians to keep abreast of the medical literature: Medical and Philosophical Commentaries, 1773-1795. Annals of Internal Medicine2000 Aug 1;133(3):238-43.
189. Simpson RJS, Pearson K. Report On Certain Enteric Fever Inoculation Statistics. The British Medical Journal1904;2(2288):1243-6.
190. Winkelstein W, Jr. The First Use of Meta-Analysis? Am J Epidemiol1998 April 15, 1998;147(8):717-.
191. Mandel H. Racial psychic history: A detailed introduction and a systematic review of investigations. Leipzig, Germany: Heims; 1936.
192. Cochrane A. Foreword. In I. Chalmers, M. Enkin,&M.J.N.C.Keirse (Eds.), Effective care in pregnancy and childbirth. Oxford, UK: Oxford University Press; 1989.
193. Chalmers I, Enkin M, Keirse M, editors. Effective care in pregnancy and childbirth. Oxford, UK: Oxford University Press; 1989.
194. Chalmers I, Altman D, editors. Systematic Reviews. 1st ed. London: BMJ Publishing Group; 1995.
195. Egger M, Smith G, Altman D. Systematic Reviews in Health Care: Meta-Analysis in Context. 2nd ed. London: BMJ Publishing Group; 2001.
196. Sandelowski M, Sandelowski M. Reading, writing and systematic review. Journal of Advanced Nursing2008 Oct;64(1):104-10.
166
197. MacLure M. Clarity bordering on stupidity: where's the quality in systematic review? Journal of Education Policy2005;20(4):393 - 416.
198. Sandelowski M, Barroso J. Handbook for Synthesizing Qualitative Research. New York Springer; 2007.
199. West S, King V, Carey T, Lohr K, McKoy N, Sutton S, et al. Systems to Rate the Strength of Scientific Evidence. Evidence Report/Technology Assessment No. 47 (Prepared by the Research Triangle Institute-University of North Carolina Evidence-based Practice Center under Contract No. 290-97-0011). AHRQ Publication No. 02-E016. Rockville, MD: Agency for Healthcare Research and Quality; 2002.
200. Ezzo J, Bausell B, Moerman DE, Berman B, Hadhazy V. Reviewing the reviews. How strong is the evidence? How clear are the conclusions? International Journal of Technology Assessment in Health Care2001;17(4):457-66.
201. Lang A, Edwards N, Fleiszer A, Lang A, Edwards N, Fleiszer A. Empty systematic reviews: hidden perils and lessons learned. Journal of Clinical Epidemiology2007 Jun;60(6):595-7.
202. Pawson R. Digging for Nuggets: How 'Bad' Research Can Yield 'Good' Evidence. International Journal of Social Research Methodology2006;9(2):127 - 42.
203. Greenland S. A Critical Look at Some Popular Meta-Analytic Methods. Am J Epidemiol1994 August 1, 1994;140(3):290-6.
204. Dickersin K, Berlin JA. Meta-analysis: state-of-the-science. Epidemiologic Reviews1992;14:154-76.
205. Shapiro S. Meta-analysis/Shmeta-analysis. Am J Epidemiol1994 November 1, 1994;140(9):771-8.
206. Eysenck H. An exercise in mega-silliness. . Am Psychol 1978;33:517.
207. Feinstein AR. Meta-analysis: statistical alchemy for the 21st century. Journal of Clinical Epidemiology1995 Jan;48(1):71-9.
208. Smith GD, Egger M. Meta-analyses of observational data should be done with due care. BMJ1999 January 2, 1999;318(7175):56-.
209. Louis TA, Fineberg HV, Mosteller F. Findings for Public Health From Meta-Analyses. Annual Review of Public Health1985;6(1):1-20.
210. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA2000 19;283(15):2008-12.
211. Egger M, Smith G, Schneider M. Systematic Reviews of Observational Studies In: Egger M, Smith GD, Altman D (eds) Systematic Reviews in Health Care: Meta-Analysis in Context. 2nd ed. London: BMJ Publishing Group; 2001.
167
212. Egger M, Schneider M, Smith GD. Meta-analysis Spurious precision? Meta-analysis of observational studies. BMJ1998 January 10, 1998;316(7125):140-4.
213. Phillips AN, Smith GD. How independent are "independent" effects? Relative risk estimation when correlated exposures are measured imprecisely. Journal of Clinical Epidemiology1991;44(11):1223-31.
214. Smith GD, Phillips AN. Confounding In Epidemiological Studies: Why "Independent" Effects May Not Be All They Seen. BMJ: British Medical Journal1992;305(6856):757-9.
215. Altman D. Systematic Reviews of Prognostic Variables In: Egger M, Smith GD, Altman D (eds) Systematic Reviews in Health Care: Meta-Analysis in Context. 2nd ed. London: BMJ Publishing Group; 2001.
216. Hemingway H. Prognosis research: Why is Dr. Lydgate still waiting? Journal of Clinical Epidemiology2006;59(12):1229-38.
217. Hayden JA, Chou R, Hogg-Johnson S, Bombardier C. Systematic reviews of low back pain prognosis had variable methods and results: guidance for future prognosis reviews. Journal of Clinical Epidemiology2009 Aug;62(8):781-96.e1.
218. Loder E, Groves T, Macauley D, Loder E, Groves T, Macauley D. Registration of observational studies. BMJ2010;340:c950.
219. Hemingway H, Riley RD, Altman DG, Hemingway H, Riley RD, Altman DG. Ten steps towards improving prognosis research. BMJ2009;339:b4184.
220. Riley RD, Ridley G, Williams K, Altman DG, Hayden J, de Vet HC, et al. Prognosis research: toward evidence-based results and a Cochrane methods group. Journal of Clinical Epidemiology2007 Aug;60(8):863-5; author reply 5-6.
221. Kotsopoulos IA, van Merode T, Kessels FG, de Krom MC, Knottnerus JA, Kotsopoulos IAW, et al. Systematic review and meta-analysis of incidence studies of epilepsy and unprovoked seizures. Epilepsia2002 Nov;43(11):1402-9.
222. Mac TL, Tran DS, Quet F, Odermatt P, Preux PM, Tan CT, et al. Epidemiology, aetiology, and clinical management of epilepsy in Asia: a systematic review. Lancet Neurology2007 Jun;6(6):533-43.
223. Burneo JG, Tellez-Zenteno J, Wiebe S, Burneo JG, Tellez-Zenteno J, Wiebe S. Understanding the burden of epilepsy in Latin America: a systematic review of its prevalence and incidence. Epilepsy Res2005 Aug-Sep;66(1-3):63-74.
224. Benamer HT, Grosset DG, Benamer HTS, Grosset DG. A systematic review of the epidemiology of epilepsy in Arab countries. Epilepsia2009 Oct;50(10):2301-4.
225. Forsgren L, Beghi E, Oun A, Sillanpaa M. The epidemiology of epilepsy in Europe - a systematic review. Eur J Neurol2005 Apr;12(4):245-53.
168
226. Preux PM, Druet-Cabanac M, Preux P-M, Druet-Cabanac M. Epidemiology and aetiology of epilepsy in sub-Saharan Africa. Lancet Neurology2005 Jan;4(1):21-31.
227. Ross SD, Estok R, Chopra S, French J. Management of Newly Diagnosed Patients with Epilepsy: A Systematic Review of the Literature. Rockville, MD: Agency for Healthcare Research and Quality (AHRQ) 2001; Available from: http://www.ncbi.nlm.nih.gov/bookshelf/br.fcgi?book=hserta&part=A56819.
228. Easterbrook PJ, Berlin JA, Gopalan R, Matthews DR. Publication bias in clinical research. Lancet1991 Apr 13;337(8746):867-72.
229. Shaheen NJ, Crosby MA, Bozymski EM, Sandler RS. Is there publication bias in the reporting of cancer risk in Barrett's esophagus? Gastroenterology2000 119(2):333-8.
230. Simunovic N, Sprague S, Bhandari M. Methodological Issues in Systematic Reviews and Meta-Analyses of Observational Studies in Orthopaedic Research. J Bone Joint Surg Am2009 May 1, 2009;91(Supplement_3):87-94.
231. Harrell FE, Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med1996 Feb 28;15(4):361-87.
232. Peduzzi P, Concato J, Feinstein AR, Holford TR. Importance of events per independent variable in proportional hazards regression analysis II. Accuracy and precision of regression estimates. Journal of Clinical Epidemiology1995;48(12):1503-10.
233. Vittinghoff E, McCulloch CE, Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol2007 Mar 15;165(6):710-8.
234. Counsell C, Dennis M. Systematic review of prognostic models in patients with acute stroke. Cerebrovascular Diseases2001;12(3):159-70.
235. Hemingway H, Marmot M. Evidence based cardiology: Psychosocial factors in the aetiology and prognosis of coronary heart disease: systematic review of prospective cohort studies. BMJ1999 May 29, 1999;318(7196):1460-7.
236. Egger M, Zellweger-Zahner T, Schneider M, Junker C, Lengeler C, Antes G. Language bias in randomised controlled trials published in English and German. Lancet1997 Aug 2;350(9074):326-9.
237. Gregoire G, Derderian F, Le Lorier J. Selecting the language of the publications included in a meta-analysis: is there a Tower of Babel bias? Journal of Clinical Epidemiology1995 Jan;48(1):159-63.
238. Moher D, Fortin P, Jadad AR, Jüni P, Klassen T, Le Lorier J, et al. Completeness of reporting of trials published in languages other than English: implications for conduct and reporting of systematic reviews. The Lancet1996;347(8998):363-6.
169
239. Moher D, Pham B, Klassen TP, Schulz KF, Berlin JA, Jadad AR, et al. What contributions do languages other than English make on the results of meta-analyses? Journal of Clinical Epidemiology2000 Sep;53(9):964-72.
240. Royle P, Bain L, Waugh N, Royle P, Bain L, Waugh N. Systematic reviews of epidemiology in diabetes: finding the evidence. BMC Medical Research Methodology2005 Jan 8;5(1):2.
241. Smith BJ, Darzins PJ, Quinn M, Heller RF. Modern methods of searching the medical literature. Medical Journal of Australia1992 Nov 2;157(9):603-11.
242. Elsevier Pharma Development Group. What are the differences between Emtree and MeSH? . 2010; Available from: http://www.info.embase.com/UserFiles/Files/Embase%20emtree_mesh_.pdf.
243. Furlan AD, Irvin E, Bombardier C, Furlan AD, Irvin E, Bombardier C. Limited search strategies were effective in finding relevant nonrandomized studies. Journal of Clinical Epidemiology2006 Dec;59(12):1303-11.
244. Wilczynski NL, Haynes RB, Hedges T, Wilczynski NL, Haynes RB. Developing optimal search strategies for detecting clinically sound prognostic studies in MEDLINE: an analytic survey. BMC Medicine2004 Jun 9;2:23.
245. Wilczynski NL, Haynes RB, Wilczynski NL, Haynes RB. Optimal search strategies for detecting clinically sound prognostic studies in EMBASE: an analytic survey. Journal of the American Medical Informatics Association2005 Jul-Aug;12(4):481-5.
246. Gøtzsche P, Harden A. Searching for non-randomized studies. In: Olsen O, Reeves B, editors. Cochrane collaboration non-randomised studies methods group Handbook. 2002.
247. Juni P, Witschi A, Bloch R, Egger M. The Hazards of Scoring the Quality of Clinical Trials for Meta-analysis. JAMA1999 September 15, 1999;282(11):1054-60.
248. Emerson JD, Burdick E, Hoaglin DC, Mosteller F, Chalmers TC. An empirical study of the possible relation of treatment differences to quality scores in controlled randomized clinical trials. Controlled Clinical Trials1990 Oct;11(5):339-52.
249. Greenland S. Quality Scores Are Useless and Potentially Misleading: Reply to "Re: A Critical Look at Some Popular Analytic Methods". Am J Epidemiol1994 August 1, 1994;140(3):300-1.
250. Wright RW, Brand RA, Dunn W, Spindler KP, Wright RW, Brand RA, et al. How to write a systematic review. Clinical Orthopaedics & Related Research2007 Feb;455:23-9.
251. Ernster VL. Nested Case-Control Studies. Preventive Medicine1994;23(5):587-90.
252. Langholz B, Richardson D, Langholz B, Richardson D. Are nested case-control studies biased? Epidemiology2009 May;20(3):321-9.
170
253. Wacholder S, Wacholder S. Bias in full cohort and nested case-control studies? Epidemiology2009 May;20(3):339-40.
254. Lubin JH, Gail MH. Biased Selection of Controls for Case-Control Analyses of Cohort Studies. Biometrics1984;40(1):63-75.
255. Mantel N. Synthetic retrospective studies and related topics. Biometrics1973 Sep;29(3):479-86.
256. Wacholder S, McLaughlin JK, Silverman DT, Mandel JS. Selection of controls in case-control studies. I. Principles. Am J Epidemiol1992 May 1;135(9):1019-28.
257. Wacholder S, Silverman DT, McLaughlin JK, Mandel JS. Selection of controls in case-control studies. III. Design options. Am J Epidemiol1992 May 1;135(9):1042-50.
258. Langholz B, Clayton D. Sampling strategies in nested case-control studies. Environmental Health Perspectives1994 Nov;102 Suppl 8:47-51.
259. Greenland S, Finkle WD. A Critical Look at Methods for Handling Missing Covariates in Epidemiologic Regression Analyses. Am J Epidemiol1995 December 15, 1995;142(12):1255-64.
260. Donders ART, van der Heijden GJMG, Stijnen T, Moons KGM. Review: A gentle introduction to imputation of missing values. Journal of Clinical Epidemiology2006;59(10):1087-91.
261. Royston P, Altman DG, Sauerbrei W, Royston P, Altman DG, Sauerbrei W. Dichotomizing continuous predictors in multiple regression: a bad idea. Stat Med2006 Jan 15;25(1):127-41.
262. Katz MH. Multivariable Analysis: A Primer for Readers of Medical Research. Annals of Internal Medicine2003 April 15, 2003;138(8):644-50.
263. Concato J, Feinstein AR, Holford TR. The Risk of Determining Risk with Multivariable Models. Annals of Internal Medicine1993 February 1, 1993;118(3):201-10.
264. Woodward M. Epidemiology: Study Design and Data Analysis 2nd ed. Boca Raton, Florida: Chapman & Hall/CRC; 2005.
265. Symons MJ, Moore DT. Hazard rate ratio and prospective epidemiological studies. Journal of Clinical Epidemiology2002 Sep;55(9):893-9.
266. Zhang J, Yu KF. What's the Relative Risk?: A Method of Correcting the Odds Ratio in Cohort Studies of Common Outcomes. JAMA1998 November 18, 1998;280(19):1690-1.
267. Viera AJMDMPH. Odds Ratios and Risk Ratios: What's the Difference and Why Does It Matter? [Miscellaneous Article]. Southern Medical Journal2008;101(7):730-4.
171
268. Devereaux PJ, Schunemann HJ, Ravindran N, Bhandari M, Garg AX, Choi PT, et al. Comparison of mortality between private for-profit and private not-for-profit hemodialysis centers: a systematic review and meta-analysis. JAMA2002 Nov 20;288(19):2449-57.
269. Warshafsky S, Lee DH, Francois LK, Nowakowski J, Nadelman RB, Wormser GP. Efficacy of antibiotic prophylaxis for the prevention of Lyme disease: an updated systematic review and meta-analysis. J Antimicrob Chemother June 1, 2010;65(6):1137-44.
270. Barth J, Schneider S, von Kanel R. Lack of Social Support in the Etiology and the Prognosis of Coronary Heart Disease: A Systematic Review and Meta-Analysis. Psychosom Med April 1, 2010;72(3):229-38.
271. Paras ML, Murad MH, Chen LP, Goranson EN, Sattler AL, Colbenson KM, et al. Sexual Abuse and Lifetime Diagnosis of Somatic Disorders: A Systematic Review and Meta-analysis. JAMA2009 August 5, 2009;302(5):550-61.
272. Holcomb WL, Jr., Chaiworapongsa T, Luke DA, Burgdorf KD. An odd measure of risk: use and misuse of the odds ratio. Obstet Gynecol2001 Oct;98(4):685-8.
273. Laupacis A, Sekar N, Stiell IG. Clinical prediction rules. A review and suggested modifications of methodological standards. JAMA1997 Feb 12;277(6):488-94.
274. Wasson JH, Sox HC, Neff RK, Goldman L. Clinical prediction rules. Applications and methodological standards. New England Journal of Medicine1985 Sep 26;313(13):793-9.
275. Royston P, Moons KG, Altman DG, Vergouwe Y, Royston P, Moons KGM, et al. Prognosis and prognostic research: Developing a prognostic model. BMJ2009;338:b604.
276. Lemeshow S, Hosmer D. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol1982;115:92 - 106.
277. Brier G. Verification of forecasts expressed in terms of probability. Monthly Weather Review1950;78:1 - 2.
278. Goldin J, Sayre JW. A guide to clinical epidemiology for radiologists: Part II statistical analysis. Clin Radiol1996 May;51(5):317-24.
279. Doust J. Using probabilistic reasoning BMJ 2009 November 3, 2009 ;339 (nov03_2): b3823.
280. Keegan M, Gali B, Findlay J, Heimbach J, Plevak D, Afessa B. APACHE III outcome prediction in patients admitted to the intensive care unit after liver transplantation: a retrospective cohort study. BMC Surgery2009;9(1):11.
281. Justice AC, Covinsky KE, Berlin JA. Assessing the generalizability of prognostic information. Annals of Internal Medicine1999 Mar 16;130(6):515-24.
172
282. Moons KG, Altman DG, Vergouwe Y, Royston P, Moons KGM, Altman DG, et al. Prognosis and prognostic research: application and impact of prognostic models in clinical practice. BMJ2009;338:b606.
283. The World Bank. World Development Indicators Washington, DC: The World Bank; 2010.
284. Marson A, Jacoby A, Johnson A, Kim L, Gamble C, Chadwick D, et al. Immediate versus deferred antiepileptic drug treatment for early epilepsy and single seizures: a randomised controlled trial. Lancet2005 Jun 11-17;365(9476):2007-13.
285. Fraser C, Murray A, Burr J. Identifying observational studies of surgical interventions in MEDLINE and EMBASE. BMC Medical Research Methodology2006;6(1):41.
286. Burnham J, Shearer B. Comparison of CINAHL, EMBASE, and MEDLINE Databases for the Nurse Researcher. Medical Reference Services Quarterly1993;12(3):45-57.
287. Woods D, Trewheellar K. Medline and Embase complement each other in literature searches. BMJ1998 April 11, 1998;316(7138):1166-.
288. Wilkins T, Gillies R, Davies K. EMBASE versus MEDLINE for family medicine searches: can MEDLINE searches find the forest or a tree? Can Fam Physician2005 June 1, 2005;51(6):848-9.
289. NCBI. Journals. 2010; Available from: http://www.ncbi.nlm.nih.gov/journals/.
290. Suarez-Almazor ME, Belseck E, Homik J, Dorgan M, Ramos-Remus C. Identifying Clinical Trials in the Medical Literature with Electronic Databases: MEDLINE Alone Is Not Enough. Controlled Clinical Trials2000;21(5):476-87.
291. Armstrong R, Jackson N, Doyle J, Waters E, Howes F. It's in your hands: the value of handsearching in conducting systematic reviews of public health interventions. J Public Health2005 December 1, 2005;27(4):388-91.
292. Bonnett L, Tudur-Smith C, Williamson P, Marson T. Prognostic factors for 12-month remission. Epilepsia2009;50:110.
293. Browne TR, Holmes GL. Epilepsy: A Primary Care Review. New England Journal of Medicine2001 April 12, 2001;344(15):1145-51.
294. Aylward RLM. Epilepsy: a review of reports, guidelines, recommendations and models for the provision of care for patients with epilepsy. Clinical Medicine, Journal of the Royal College of Physicians2008;8:433-8.
295. Peleg R, Shvartzman P. Where Should Family Medicine Papers be Published--Following the Impact Factor? J Am Board Fam Med2006 November 1, 2006;19(6):633-6.
296. Altman DG, Altman DG. Prognostic models: a methodological framework and review of models for breast cancer. Cancer Investigation2009 Mar;27(3):235-43.
173
297. Pocock SJ, Collier TJ, Dandreo KJ, de Stavola BL, Goldman MB, Kalish LA, et al. Issues in the reporting of epidemiological studies: a survey of recent practice. BMJ2004 October 16, 2004;329(7471):883-.
298. Vandenbroucke JP, Elm Ev, Altman DG, Gøtzsche PC, Mulrow CD, Pocock SJ, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): Explanation and Elaboration. Annals of Internal Medicine2007 October 16, 2007;147(8):W-163-W-94.
299. STROBE. The STROBE Statement (Strengthening the Reporting of Observational Studies in Epidemiology) Endorsement. ISPM, University of Bern 2009; 2009; Available from: http://www.strobe-statement.org/index.php?id=strobe-endorsement.
300. Woolfenden S, Ridley G, Williams K. Developing a classification system for a systematic review of the prognosis of autism Cochrane Colloquium Abstracts Cochrane Collaboration; 2007.
301. Elwes RDC, Johnson AL, Reynolds EH. The Course Of Untreated Epilepsy. BMJ: British Medical Journal1988;297(6654):948-50.
302. Reynolds EH. Do anticonvulsants alter the natural course of epilepsy? Treatment should be started as early as possible. BMJ1995 Jan 21;310(6973):176-7.
303. Nikanfar M, Arami MA, Mansourpoor L, Najmi S. Common etiologies of adult onset epilepsies in northwest of Iran. Acta Medica Iranica2005;43(3):223-6.
304. Stephen LJ, Brodie MJ. Epilepsy in elderly people. Lancet2000;355(9213):1441-6.
305. Shorvon S, Perucca E, Fish D, Dodson W, editors. The Treatement of Epilepsy. 2nd ed. Oxford: Blackwell; 2004.
306. Diamond GA. Future imperfect: the limitations of clinical prediction models and the limits of clinical prediction. J Am Coll Cardiol1989 Sep;14(3 Suppl A):12A-22A.
307. Shrier I, Steele R, Shrier I, Steele R. Understanding the relationship between risks and odds ratios. Clin J Sport Med2006 Mar;16(2):107-10.
308. McNutt LA, Wu C, Xue X, Hafner JP, McNutt L-A, Wu C, et al. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol2003 May 15;157(10):940-3.
309. Barros A, Hirakata V. Alternatives for logistic regression in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Medical Research Methodology2003;3(1):21.
310. Wacholder S. Binomial regression in glim: estimating risk ratios and risk differences. Am J Epidemiol1986 January 1, 1986;123(1):174-84.